Vehicle-mounted sleep-aiding sound source dynamic adjustment system and method based on brain wave perception

Abstract

The invention relates to the field of intelligent cabin technology and biological signal processing, in particular to a vehicle-mounted sleep-aiding sound source dynamic adjustment system and method based on brain wave perception. The system comprises a brain wave signal acquisition filtering module which is used for preprocessing acquired brain wave signals and generating a pulse sequence; the SNN simulation processing module is used for inputting the pulse sequence into a pre-established pulse neural network SNN model for brain-like simulation calculation and identifying a current sleep stage; the vehicle machine sleep-aiding module is used for selecting and generating a corresponding sleep-aiding stimulation signal from a preset sound source library according to the identified current sleep stage; and the headrest player is used for playing a corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal. According to the scheme, dynamic decision making is carried out by integrating multi-dimensional data, the single judgment logic only depending on the distance is broken, the automatic ending function is real and reliable, the driving safety is improved, and the driving risk caused by distracted operation is effectively reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent cockpit technology and biological signal processing, and particularly relates to a vehicle-mounted sleep-aiding sound source dynamic adjustment system and method based on brain wave perception. BACKGROUND

[0002] With the acceleration of social rhythm and the increase of work pressure, people's demand for resting in the car during the commute gap or long-distance travel is increasingly evident. However, factors such as the small space in the car, environmental noise interference, and limited seat comfort often lead to poor rest quality and cannot effectively relieve fatigue. To improve comfort, existing intelligent cars generally use ergonomic seats, heating, ventilation, and massage functions, ambient lights, and other technical solutions. However, these solutions mostly optimize the driving experience in a conscious state and still lack sufficient support for the specific physiological state of sleep.

[0003] There are some vehicle-mounted sleep assistance solutions in the prior art. For example, patent application CN202210506855.6 discloses a system that helps sleep by adjusting environmental factors such as temperature, humidity, and air volume in the car. However, this method belongs to passive environmental adjustment and lacks precision and adaptability as it does not start from the user's physiological state. Another patent application CN202410313942.9 mentions assisting passengers to sleep, but does not involve real-time perception and feedback control based on physiological signals, especially does not solve the challenge of real-time electroencephalogram (EEG) processing in a low-power vehicle-mounted environment.

[0004] Electroencephalogram (EEG) is the most direct indicator of brain neural electrical activity. The brain waves in different sleep stages (such as wakefulness, rapid eye movement (REM), light sleep, and deep sleep) have significant differences in frequency and pattern, which provides a medical theoretical basis for accurately identifying sleep stages. Traditional EEG recognition algorithms are mostly based on convolutional neural networks (CNN) or recurrent neural networks (RNN). Although these methods are effective, they have high computational complexity and high power consumption, making it difficult to achieve real-time processing in vehicle-mounted embedded systems.

[0005] Therefore, there is an urgent need in the art for a vehicle-mounted system that can accurately and efficiently identify sleep stages in real time and dynamically provide personalized sleep-aiding stimuli based on the identification results to truly improve the rest quality of users in the car. SUMMARY

[0006] In order to overcome the deficiencies in the prior art, the present application provides a vehicle-mounted sleep-aiding sound source dynamic adjustment system and method based on brain wave perception, aiming to solve the problem that the existing vehicle-mounted rest system cannot provide accurate sleep-aiding services based on the real-time physiological state of the user. A spiking neural network (SNN) is used for brain-like simulation calculation to realize real-time and accurate identification of sleep stages with extremely low power consumption, overcoming the power consumption and computing power bottleneck of traditional deep learning models deployed on vehicle-mounted embedded platforms. According to the identified sleep stage, different characteristic sound sources are adaptively played to scientifically guide the user's sleep process from shallow to deep and improve the rest efficiency.

[0007] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0008] In a first aspect, the present application provides a vehicle-mounted sleep-aiding sound source dynamic adjustment system based on brain wave perception, which comprises:

[0009] A brain wave signal acquisition and filtering module is used to preprocess the acquired electroencephalogram signals to generate a pulse sequence.

[0010] An SNN simulation processing module is used to input the pulse sequence into a pre-established spiking neural network (SNN) model for brain-like simulation calculation to identify the current sleep stage.

[0011] A car machine sleep-aiding module is used to select and generate a sleep-aiding stimulation signal corresponding to the identified current sleep stage from a pre-set sound source library.

[0012] A headrest player is used to play the corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal.

[0013] Optionally, the brain wave signal acquisition module comprises a vehicle-mounted sleep-aiding cap, and a pulse discharge filter and a pulse encoding submodule arranged on the seat backrest in the vehicle.

[0014] The vehicle-mounted sleep-aiding cap is used to acquire the electroencephalogram signals of the driver or passenger.

[0015] The pulse discharge filter is used to filter and process the electroencephalogram signals.

[0016] A pulse encoding submodule is used to pulse encode the filtered and processed electroencephalogram signals to generate a pulse sequence.

[0017] Optionally, the pulse encoding submodule comprises:

[0018] An extraction unit is used to extract the power values of different frequency bands in the filtered and processed electroencephalogram signals.

[0019] A processing unit is used to normalize the power values, and based on the normalized power values, a pulse sequence corresponding to each frequency band is generated using a firing frequency encoding method.

[0020] Optionally, the SNN simulation processing module includes: a spiking neural network (SNN) model; the SNN model is trained using the STDP rule to adjust the synaptic connection weights between neurons; if the presynaptic neuron fires a pulse before the postsynaptic neuron, the synaptic weight is increased; if the presynaptic neuron fires a pulse after the postsynaptic neuron, the synaptic weight is decreased.

[0021] Optionally, the spiking neural network (SNN) model includes at least one hidden layer, an input layer, and an output layer;

[0022] The input layer is used to receive pulse sequences;

[0023] The hidden layer consists of two network layers; each network layer includes several sparsely connected LIF neurons for performing brain-like simulation calculations.

[0024] The output layer is used to classify the brain-like simulation calculation results using a Softmax classifier, output the probability of the user being in different sleep stages, and determine the sleep stage with the highest probability as the current sleep stage.

[0025] The sleep stages include wakefulness, REM sleep, light sleep, and deep sleep.

[0026] Optionally, the in-vehicle sleep aid module includes:

[0027] The first selection unit is used to select, from a preset sound source library, a source containing alpha wave induction if the current sleep stage is during wakefulness.

[0028] The sound source signal of the component;

[0029] The second selection unit is used to select a sound source signal containing delta wave-induced components from a preset sound source library if the current sleep stage is REM sleep.

[0030] The third selection unit is used to select a sound source signal containing a specific frequency sleep spindle wave enhancement component from a preset sound source library if the current sleep stage is light sleep.

[0031] The fourth selection unit is used to select a low-intensity, stable-frequency sound source signal from a preset sound source library if the current sleep stage is deep sleep.

[0032] The generation unit is used to dynamically adjust the sound source type and parameters of different sound source signals according to the identified sleep stage, and generate a sleep-aiding stimulation signal that matches the current sleep stage.

[0033] Optionally, the in-vehicle sleep aid module further includes:

[0034] The sleep quality evaluation unit is configured to record sleep data of a current sleep stage, combine historical sleep data of the same driver or passenger in different sleep stages, and evaluate sleep quality.

[0035] The data synchronization unit is configured to synchronize the sleep data to an external health device.

[0036] In a second aspect, the present application provides a vehicle-mounted sleep-aiding sound source dynamic adjustment method based on brain wave perception, which comprises the following steps:

[0037] The collected brain electrical signals are preprocessed to generate a pulse sequence.

[0038] The pulse sequence is input into a pre-established spiking neural network (SNN) model for brain-like simulation calculation to identify a current sleep stage.

[0039] According to the identified current sleep stage, a sleep-aiding stimulation signal corresponding thereto is selected and generated from a pre-set sound source library.

[0040] The corresponding sleep-aiding sound source is played according to the sleep-aiding stimulation signal.

[0041] In a third aspect, the present application provides an electronic device, which comprises:

[0042] at least one processor; and

[0043] a memory in communication with the at least one processor; wherein

[0044] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the method of the second aspect.

[0045] Compared with the closest prior art, the present application has the beneficial effects that:

[0046] The vehicle-mounted sleep-aiding sound source dynamic adjustment system based on brain wave perception comprises a brain wave signal acquisition and filtering module, an SNN simulation processing module, a vehicle-machine sleep-aiding module, and a headrest player.

[0047] The core calculation of the application adopts a spiking neural network (SNN), as a brain-computer model driven by events, the calculation activity only occurs when a pulse is received, compared with a traditional deep neural network (DNN / CNN), the calculation power consumption is significantly reduced, so that it is possible to realize real-time and continuous electroencephalogram signal processing on a vehicle embedded MCU.

[0048] The application scheme of the application forms a closed loop feedback of 'perception-identification-decision-stimulation'. The sleep state change of the user can be dynamically tracked, and the sleep assisting strategy can be adaptively adjusted, so that personalized sleep guiding service is provided, and the user experience is improved.

[0049] The application ingeniously integrates electroencephalogram acquisition, calculation processing, decision interaction and headrest loudspeaker into the whole vehicle intelligent cabin ecology, improves the technological added value and market competitiveness of the vehicle. The sleep data recorded by the system can be synchronized with external devices such as the health bracelet of the user, forming a personal health record, and providing data support for subsequent health management service. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the specific embodiments of the application or the technical solutions in the prior art, the drawings needed to be used in the specific embodiments or the prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual proportion.

[0051] Figure 1 is a working framework schematic diagram of a vehicle-mounted sleep assisting sound source dynamic adjustment method based on electroencephalogram perception provided by the application;

[0052] Figure 2 is a whole structure schematic diagram of a vehicle-mounted sleep assisting sound source dynamic adjustment system based on electroencephalogram perception provided by the application;

[0053] Figure 3 is a flowchart of a vehicle-mounted sleep assisting sound source dynamic adjustment method based on electroencephalogram perception provided by the application;

[0054] Figure 4 is an internal structure diagram of an electronic device provided by the application. DETAILED DESCRIPTION

[0055] The embodiments of the technical solutions of the application will be described in detail below with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the application, and therefore only serve as examples, and cannot limit the protection scope of the application.

[0056] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in the present application should be understood as the usual meanings understood by the skilled person in the field of the application.

[0057] The application provides a brain wave perception-based vehicle-mounted sleep-aiding sound source dynamic adjustment system and method. In particular, it relates to a brain-computer interface (BCI) and spiking neural network (SNN) combined brain-like computing system. Specifically, it is a vehicle-mounted sleep-aiding system and method based on electroencephalogram (EEG) intelligent perception, which can identify the sleep stage of the driver or passenger in real time and dynamically adjust the sleep-aiding sound source. The method and system can improve the riding experience of the intelligent cabin, reduce fatigue when the driver or passenger is sleeping, increase the configuration of the intelligent cabin, improve the configuration value of the vehicle, and improve the vehicle service. The embodiments of the application will be described below with reference to the accompanying drawings.

[0058] Embodiment 1: Please refer to Figure 1 and 2 Embodiment 1 of the application provides a brain wave perception-based vehicle-mounted sleep-aiding sound source dynamic adjustment system, as shown in Figure 2 , which comprises a brain wave signal acquisition and filtering module 210, an SNN simulation processing module 220, a vehicle-mounted sleep-aiding module 230, and a headrest player 240, wherein:

[0059] The brain wave signal acquisition and filtering module 210 is used to preprocess the collected electroencephalogram signal to generate a pulse sequence.

[0060] The SNN simulation processing module 220 is used to input the pulse sequence into a pre-established spiking neural network (SNN) model for brain-like simulation calculation to identify the current sleep stage.

[0061] The vehicle-mounted sleep-aiding module 230 is used to select and generate a sleep-aiding stimulation signal corresponding to the identified current sleep stage from a pre-set sound source library.

[0062] The headrest player 240 is used to play the corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal.

[0063] In the above embodiment, the brain wave signal acquisition module comprises a vehicle-mounted sleep-aiding cap, and a pulse discharge filter and a pulse encoding submodule arranged on the seat backrest in the vehicle.

[0064] The vehicle-mounted sleep-aiding cap is used to collect the electroencephalogram signal of the driver or passenger.

[0065] The pulse discharge filter is used to filter the electroencephalogram signal.

[0066] The pulse encoding submodule is used to pulse encode the filtered electroencephalogram signal to generate a pulse sequence.

[0067] In the above embodiment, the pulse encoding submodule comprises:

[0068] An extraction unit is configured to extract power values of different frequency bands in the filtered electroencephalogram signal.

[0069] A processing unit is configured to normalize the power values, and generate pulse sequences corresponding to the frequency bands based on the normalized power values by using a firing rate coding method.

[0070] In one embodiment, in the above embodiment, the seat headrest is configured with a headrest sleep-aiding player; the seat back is configured with a pulse discharge filter inside; and the seat is configured with an SNN simulation processing module (a microprocessor MCU for SNN calculation, which is also a microcomputer, for pulse neural network calculation, and can also be used for data communication with the vehicle machine).

[0071] The seat is configured with an external accessory: a sleep-aiding cap (a portable cap for collecting electroencephalogram).

[0072] The vehicle machine and the SNN simulation processing module are connected through WIFI and Bluetooth, and are used for transmitting brainwave classification and recognition result coding data and sleep stage recognition data.

[0073] The vehicle machine system is installed with a sleep-aiding application, which can use sleep data to evaluate sleep quality, and can synchronize sleep data with a health bracelet.

[0074] When a driver or passenger sleeps in the intelligent cabin, the sleep-aiding headgear can be worn. The sleep-aiding headgear collects electroencephalogram voltage signals, filters them into pulse signals for SNN neural network calculation through a filter. After the SNN neural network, the sleep brainwaves of a specific stage are simulated and classified into specific brainwave patterns by the SNN simulation. After the classified brainwave patterns are compressed, they are transmitted to the vehicle machine through Bluetooth and WIFI (Bluetooth connection and WIFI data transmission, i.e. air transmission) to determine the sleep stage type. After the vehicle machine records the sleep data, it can synchronize the data to the health bracelet, and can also air transmit the sleep type of the current driver or passenger to the SNN simulation processing module to generate sleep-aiding stimulation signals (sleep-aiding sound source signals). Finally, the sleep-aiding sound source signals are transmitted to the headrest player, which plays the corresponding sound source for stimulating sleep.

[0075] In the above embodiment, the SNN simulation processing module comprises a pulse neural network SNN model; the pulse neural network SNN model is trained by using an STDP rule to adjust the synaptic connection weight between neurons; if a presynaptic neuron fires a pulse before a postsynaptic neuron, the synaptic weight is increased; if a presynaptic neuron fires a pulse after a postsynaptic neuron, the synaptic weight is decreased.

[0076] In the above embodiment, the pulse neural network SNN model comprises at least one hidden layer, an input layer and an output layer.

[0077] The input layer is configured to receive the pulse sequence;

[0078] The hidden layer is composed of two layers of networks, each of which includes a plurality of LIF neurons connected sparsely and configured to perform brain simulation calculation;

[0079] The output layer is configured to classify the brain simulation calculation result by using a Softmax classifier, output probabilities of the user being in different sleep stages, and determine the sleep stage with the highest probability as the current sleep stage.

[0080] The sleep stages include a wake stage, a rapid eye movement sleep stage, a light sleep stage, and a deep sleep stage.

[0081] Specifically, the SNN processes the electroencephalogram signal; the SNN is different from the traditional neural network, which uses pulses (discrete time events) instead of only continuous activation values to transmit information. The activation of neurons is represented by "spiking" events, and the time scale of spiking is often part of the information.

[0082] Input data: The original EEG signal collected by the electroencephalogram cap may contain noise and needs to be preprocessed.

[0083] [Preprocessing] Denoising, using environmental noise of about 50 Hz (also 60 Hz) (the noise frequency varies depending on the material of the device, which can be measured by "empty collection").

[0084] [Pulse coding] The actual coding of the electroencephalogram is feature extraction of the electroencephalogram. According to sleep medicine theory, each stage of sleep presents obvious classifiable electroencephalogram patterns:

[0085] Rapid eye movement sleep: The electroencephalogram of the REM stage is similar to the wake state or the lighter light sleep, often showing mixed frequency electroencephalogram activity. It includes low-amplitude, high-frequency beta waves (13-30 Hz) and some theta waves.

[0086] Light sleep: NREM stages 1 and 2.

[0087] NREM stage 1: This is the light sleep stage just after falling asleep. The electroencephalogram shows that alpha waves (8-13 Hz) gradually decrease and are replaced by theta waves (4-7 Hz).

[0088] NREM stage 2: Slightly deeper than NREM stage 1. The electroencephalogram shows that theta waves are dominant, but sleep spindles (12-16 Hz) and K-complexes (large-amplitude negative waves followed by positive waves) also appear, which are typical markers of NREM stage 2.

[0089] Deep sleep: Stage 3 of NREM. In deep sleep stage, the electroencephalogram mainly shows delta waves (frequency less than 4 Hz, high amplitude slow wave).

[0090] According to the theory of sleep medicine: different sleep stages correspond to different frequency wave differences, which can be used as identification basis. Therefore, in the encoding stage, the power values of different frequency bands (θ, δ, α, β) are calculated, and the "firing rate coding" method (this method can normalize the continuous feature values into pulse sequences) is used to generate pulse sequences (4 sequences corresponding to θ, δ, α, β) in each unit time (1 min) as the input of SNN neural network.

[0091] Network structure:

[0092] Input layer: a total of 4 input neurons, receiving pulse signal values (0, 1 sequence, 1 indicating firing pulse, 0 indicating resting).

[0093] Hidden layer: a total of 2 layers of network, each layer including 20 LIF neurons, using sparse connection between layers. LIF neuron is a biological simulation neuron, realizing brain simulation calculation.

[0094] LIF neuron excitation: LIF neuron receives a pulse every time, and the dynamic model will emit a simulation graph in the form of the following figure.

[0095] LIF neuron dynamic model:

[0096]

[0097] Where, V(t) is the membrane potential (voltage value) of the neuron at time t; V rest is the resting potential, which can be set to 0. I(t) is the incoming synaptic current, τ is the time constant; R is the membrane resistance.

[0098] Input current calculation: the input current I(t) comes from the pulse firing of the previous layer of neurons, and the weight w and synaptic delay are combined together.

[0099] Where, w ij is the synaptic weight from the previous layer of neurons j to the current neuron i, is the pulse response of the synapse, commonly using an exponential decay function.

[0100] Sparse connection implementation: In the network construction stage, full connection is used between two layers of neurons in the hidden layer. A connection probability constant parameter P (0<=P<1) is set between the connections and applied to the second layer of neurons. P represents the pre-connection probability of this neuron. When P == 0, it is equivalent to no connection, and the P value can be adjusted according to the recognition result effect.

[0101] Output layer: The output layer uses the softmax method. The hidden layer and the softmax layer are connected in a full connection manner. Because the sleep categories need to be classified, i.e., three categories of rapid eye movement, light sleep, and deep sleep stage, this layer uses three neurons. The difference between the softmax neuron and the LIF neuron is that the LIF neuron can count the pulses in a unit time, and by calculating the pattern score of the hidden layer neuron output, the probability that the pattern score belongs to the classified category is obtained.

[0102] Softmax formula of K categories: , x is the activation value (pattern score) of the neuron.

[0103] Pattern score output by the first softmax neuron: .

[0104] Where, represents the pulse count in a unit time (1 min).

[0105] Calculation period: one round of simulation operation is performed every 1 min to obtain the sleep category probability in 1 min, and the category with the maximum probability is found.

[0106] Learning rule: STDP (Spike-Timing-Dependent Plasticity) weight learning rule is adopted.

[0107] Specifically, if a presynaptic neuron fires a pulse before a postsynaptic neuron, the synaptic weight increases (LTP, Long-Term Potentiation, weighting)

[0108] If a presynaptic neuron fires a pulse after a postsynaptic neuron, the synaptic weight decreases (LTD, Long-Term Depression, weighting reduction).

[0109] In the above embodiment, the car machine sleep aid module comprises:

[0110] The first selection unit is configured to select a sound source signal containing an alpha wave induction

[0111] component from a preset sound source library if the current sleep stage is in the wake-up period.

[0112] The second selection unit is configured to select a sound source signal containing a delta wave inducing component from a preset sound source library if the current sleep stage is in the rapid eye movement sleep period.

[0113] The third selection unit is configured to select a sound source signal containing a specific frequency sleep spindle wave enhancing component from the preset sound source library if the current sleep stage is in the light sleep period.

[0114] The fourth selection unit is configured to select a sound source signal playing a low-intensity and stable frequency from the preset sound source library if the current sleep stage is in the deep sleep period.

[0115] The generation unit is configured to dynamically adjust the sound source type and parameters of different sound source signals according to the identified sleep stage, and generate a sleep-aiding stimulation signal matching the current sleep stage.

[0116] In the above embodiment, the vehicle machine sleep-aiding module further comprises:

[0117] The sleep quality evaluation unit is configured to record sleep data of the current sleep stage, and evaluate the sleep quality by combining historical sleep data of the same driver or passenger in different sleep stages.

[0118] The data synchronization unit is configured to synchronize the sleep data to an external health device.

[0119] Embodiment 2: In order to further illustrate the working principle of the system in Embodiment 1, Embodiment 2 of the present application provides a vehicle-mounted sleep-aiding sound source dynamic adjustment system based on brain wave perception, and the overall architecture is as shown in Figure 1

[0120] The brain wave signal acquisition and filtering module comprises a portable vehicle-mounted sleep-aiding cap (which can be internally provided with dry electrodes or micro-wet electrodes) for acquiring original electroencephalogram (EEG) of a user. The signal is transmitted to a pulse discharge filter located in the seat backrest through a wired or low-frequency Bluetooth mode. The filter first pre-processes the signal, and the core is to filter out 50Hz (or 60Hz) power frequency interference using a band-stop filter (Notch Filter). Subsequently, the signal enters a pulse coding sub-module.

[0121] Referring to Figure 1 ​The pulse coding process is as follows: the extraction unit in the coding submodule performs short-time Fourier transform (STFT) on the filtered EEG signal or uses a set of band-pass filters to extract the power values of the four classic frequency bands of delta (0.5-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), and beta (13-30 Hz). The processing unit then normalizes the four power values to the range of [0, 1] respectively, and then uses the firing rate coding method to convert each normalized power value into a pulse sequence. The conversion rule is: in a unit time T (for example, 1 minute), the higher the normalized value, the more pulses (represented by '1') generated, and the less rest (represented by '0'). Finally, four parallel pulse sequences are generated as the input of the SNN.

[0122] The SNN simulation processing module is deployed on a low-power microprocessor (MCU) under the seat. The module runs a pre-trained pulse neural network model.

[0123] The pulse neural network SNN model includes:

[0124] Input layer: composed of 4 neurons, respectively receiving the above 4 pulse sequences.

[0125] Hidden layer: composed of 2 layers of networks, each layer containing 20 LIF (Leaky Integrate-and-Fire) neurons. LIF neurons are highly biologically simulated models, whose dynamic characteristics are described by the differential equation:

[0126]

[0127] Where V(t) is the membrane potential (voltage value) of the neuron at time t; V rest is the resting potential, which can be set to 0. I(t) is the incoming synaptic current, τ is the time constant; R is the membrane resistance.

[0128] The neuron connections between layers are not fully connected, but a connection probability P (for example, P=0.6) is set to achieve sparse connection, further reducing the computational load and overfitting risk.

[0129] Output layer: contains 3 neurons, respectively corresponding to "REM sleep", "light sleep (N1+N2)", and "deep sleep (N3)". The output layer uses the Softmax function, which normalizes the total number of pulses (i.e., pulse count) fired by the hidden layer neurons in a unit time window, outputs the probability of belonging to each sleep stage, and takes the class with the highest probability as the final recognition result.

[0130] The SNN model is trained using the STDP (Spike-Timing-Dependent Plasticity) rule (offline training mode). STDP is an unsupervised biologically-inspired learning rule: if the pre-synaptic neuron fires earlier than the post-synaptic neuron (i.e., Δt = t_post - t_pre > 0), the synaptic weight is increased (LTP); otherwise, the weight is decreased (LTD). The weight change roughly follows an exponential decay law. In actual training, we use a supervised STDP method with a teacher signal: when the network output is wrong, a strong pulse is injected into the correct output neuron to induce it to fire, thereby using the STDP rule to perform error backpropagation and weight adjustment throughout the network.

[0131] The car machine sleep aid module runs in the form of an APP on the car machine system. It receives sleep stage recognition results from the SNN module through Wi-Fi. Its internal logic is as follows:

[0132] If it is identified as a wake-up period, the first selection unit selects a sound source containing an alpha wave (8-13 Hz) inducing component (such as a soothing natural wind sound, stream sound) from the sound source library, aiming to help the user relax and enter a sleep preparation state.

[0133] If it is identified as a rapid eye movement sleep period (REM), the second selection unit selects a sound source containing a delta wave (0.5-4 Hz) inducing component (such as deep, slow beats), aiming to guide the user into deeper non-rapid eye movement sleep.

[0134] If it is identified as a light sleep period (NREM N1 / N2), the third selection unit selects a sound source containing a specific frequency (12-16 Hz) sleep spindle enhancement component, aiming to stabilize sleep, reduce the number of awakenings, and promote the transition to deep sleep.

[0135] If it is identified as a deep sleep period (NREM N3), the fourth selection unit selects a low-intensity, stable-frequency maintenance sound source (such as uniform white noise), aiming to avoid disturbing the user and protecting their deep sleep state.

[0136] The generation unit is responsible for generating the final audio signal and ensuring smooth transitions in volume and frequency of the sound source when switching between different sleep stages, avoiding abrupt changes that may wake up the user.

[0137] In addition, the sleep quality assessment unit of this module records the duration of each stage, the number of transitions, and other data, and comprehensively assesses the sleep quality of this rest. The data synchronization unit can synchronize the sleep report to the user's Huawei, Xiaomi, and other health bands through Bluetooth.

[0138] The headrest player receives the audio signal sent by the car machine and converts it into sound waves to play to the user, completing the entire sleep-aiding closed loop.

[0139] Based on the same inventive concept, the embodiments of the present application also provide a brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment method for implementing the above-mentioned brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment system. The implementation scheme for solving the problem provided by the method is similar to the implementation scheme described in the above-mentioned system embodiment, so the specific limitations in one or more brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment method embodiments provided below can be referred to the limitations of the brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment system described above, which will not be repeated here.

[0140] Embodiment 3: Please refer to Figure 3 , Figure 3 A brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment method is provided for the embodiment 1 of the present application, which can be applied to a vehicle-mounted system. The method specifically includes the following steps:

[0141] S101: Preprocessing the collected brain electrical signals to generate a pulse sequence;

[0142] S102: Inputting the pulse sequence into a pre-established spiking neural network (SNN) model for brain-like simulation calculation to identify the current sleep stage;

[0143] S103: According to the identified current sleep stage, selecting and generating a sleep-aiding stimulation signal corresponding thereto from a pre-set sound source library;

[0144] S104: Playing a corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal.

[0145] Embodiment 4: In order to further illustrate the above steps, the embodiment 4 of the present application simulates the actual use scenario of the present application scheme and provides a brain wave perception-based vehicle sleep-aiding sound source dynamic adjustment method, which includes the following steps:

[0146] S1: The user wears a sleep-aiding cap, and the system is powered on.

[0147] S2: The brain wave signal acquisition and filtering module continuously acquires and processes the brain electrical signals to generate a pulse sequence.

[0148] S3: The SNN simulation processing module receives the pulse sequence and performs brain-like simulation calculation.

[0149] S4: The SNN model outputs the identification result of the current sleep stage.

[0150] S5: The car machine sleep-aiding module selects a matching sound source type from the sound source library according to the identification result.

[0151] S6: The car machine sleep-aid module generates sleep-aid stimulation signals and sends them to the headrest player.

[0152] S7: The headrest player plays the sleep-aid sound source.

[0153] S8: The system determines whether the current break is over (for example, the user takes off the hat or the vehicle starts). If not, return to step S602 for the next round of perception and adjustment; if so, proceed to S9.

[0154] S9: The car machine sleep-aid module generates a sleep report and synchronizes it to an external health device, and the process ends.

[0155] In one embodiment, an electronic device, which can be a terminal, has an internal structure diagram as shown in Figure 4 The electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the electronic device is used for wired or wireless communication with external terminals. Wireless communication can be achieved through WIFI, mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program is executed by the processor to implement any one of steps S101 to S104 of the brain wave perception-based dynamic adjustment method of the vehicle-mounted sleep-aid sound source. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad provided on the shell of the electronic device. It can also be an external keyboard, touchpad, or mouse, etc.

[0156] Those skilled in the art can understand that Figure 4 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. A specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0157] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In one embodiment, the application can be implemented in software and can be stored on a computer readable medium, which can include random access memory (RAM), read only memory (ROM), magnetic disk or optical disk, or the like. The application can also be implemented as a combination of both software and hardware. Embodiments of the application can be implemented as a method, apparatus, or computer program product using programming or design tools, which are both well-known and available in the art.

[0158] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0159] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0160] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0161] The foregoing is merely illustrative of the principles of this application and various modifications can be made by those skilled in the art without departing from the scope of the application. The above-described embodiments are presented for purposes of illustration and for conveying the principles of the application. It is understood that the application can be practiced with modifications and alterations.

Claims

1. A brain wave perception-based vehicle-mounted sleep-aiding sound source dynamic adjustment system, characterized in that, The method comprises: A brain wave signal collection filtering module is configured to pre-process the collected electroencephalogram signals to generate a pulse sequence; An SNN simulation processing module is configured to input the pulse sequence into a pre-established spiking neural network (SNN) model to perform brain-like simulation calculation and identify a current sleep stage; A car-machine sleep-aiding module is configured to select and generate a sleep-aiding stimulation signal corresponding to the identified current sleep stage from a pre-set sound source library; A headrest player is configured to play a corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal.

2. The system of claim 1, wherein, The brain wave signal collection module comprises a vehicle-mounted sleep-aiding cap, a pulse discharge filter and a pulse coding submodule arranged on a seat backrest in a vehicle; The vehicle-mounted sleep-aiding cap is configured to collect electroencephalogram signals of a driver or a passenger; The pulse discharge filter is configured to filter the electroencephalogram signals; The pulse coding submodule is configured to pulse code the filtered electroencephalogram signals to generate a pulse sequence.

3. The system of claim 2, wherein, The pulse coding submodule comprises: An extraction unit is configured to extract power values of different frequency bands in the filtered electroencephalogram signals; A processing unit is configured to normalize the power values, generate pulse sequences corresponding to each frequency band based on the normalized power values, and adopt a firing rate coding method.

4. The system of claim 1, wherein, The SNN simulation processing module comprises an SNN model; the SNN model is trained using an STDP rule to adjust synaptic connection weights between neurons; if a presynaptic neuron fires a pulse before a postsynaptic neuron, the synaptic weight is increased; if the presynaptic neuron fires a pulse after the postsynaptic neuron, the synaptic weight is decreased.

5. The system of claim 4, wherein, The SNN model comprises at least one hidden layer, an input layer and an output layer; The input layer is configured to receive the pulse sequence; The hidden layer is composed of two layers of networks; each layer of network comprises a plurality of LIF neurons connected sparsely and configured to perform brain-like simulation calculation; The output layer is configured to classify the brain-like simulation calculation results using a Softmax classifier, output probabilities of the user being in different sleep stages, and determine the sleep stage with the highest probability as the current sleep stage; The sleep stages include a wake period, a rapid eye movement sleep period, a light sleep period and a deep sleep period.

6. The system of claim 1, wherein, The car-machine sleep-aiding module comprises: A first selection unit is configured to select a sound source signal containing an alpha wave induction component from a pre-set sound source library if the current sleep stage is in the wake period; A second selection unit is configured to select a sound source signal containing a delta wave induction component from the pre-set sound source library if the current sleep stage is in the rapid eye movement sleep period; A third selection unit is configured to select a sound source signal containing a specific frequency sleep spindle wave enhancement component from the pre-set sound source library if the current sleep stage is in the light sleep period; A fourth selection unit is configured to select a sound source signal playing a low-intensity, stable-frequency sound from the pre-set sound source library if the current sleep stage is in the deep sleep period; A generation unit is configured to dynamically adjust sound source types and parameters of different sound source signals according to the identified sleep stage, and generate a sleep-aiding stimulation signal matching the current sleep stage. The car-machine sleep-aiding module further comprises:

7. The system of claim 6, wherein, ​ The sleep quality evaluation unit is configured to record sleep data of a current sleep stage, combine historical sleep data of the same driver or passenger in different sleep stages, and evaluate sleep quality. The data synchronization unit is configured to synchronize the sleep data to an external health device.

8. A brain wave perception-based dynamic adjustment method for a vehicle-mounted sleep-aiding sound source, characterized in that, The method comprises: Preprocessing the collected electroencephalogram signals to generate a pulse sequence; Inputting the pulse sequence into a pre-established spiking neural network (SNN) model for brain-like simulation calculation to identify a current sleep stage; According to the identified current sleep stage, selecting and generating a sleep-aiding stimulation signal corresponding thereto from a pre-set sound source library; Playing a corresponding sleep-aiding sound source according to the sleep-aiding stimulation signal.

9. An electronic device, comprising: The electronic device comprises: At least one processor; and A memory connected in communication with the at least one processor; wherein The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the method of claim 8.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method of claim 8.

Images