Intelligent home control system combined with brain-computer interface

Abstract

The invention discloses an intelligent home control system combined with a brain-computer interface, which comprises an electroencephalogram signal acquisition device, a signal processing unit and an intelligent home control unit, and is characterized in that the electroencephalogram signal acquisition device is connected with the signal processing unit, and the signal processing unit is connected with the intelligent home control unit; and the signal processing unit is also connected with a cloud server. The brain-computer interface technology is adopted to analyze the brain activity of the user in real time, the requirements and intentions of the user can be known, the purpose of controlling the home equipment through the idea is achieved, the interaction mode is more natural and visual, manual operation is not needed, unprecedented convenient experience is provided for the user, and the intelligent home control system has great significance for old people and handicapped people who are unable to move freely. Therefore, the users can control their own houses more independently, and the self-care ability of life is improved.

Description

Technical Field

[0001] The present invention relates to the field of smart home control technology, and in particular to a smart home control system combined with a brain-computer interface. Background Art

[0002] With the continuous advancement of technology, people's demand for smart homes is also growing. Traditional smart home control methods mainly rely on manual operation or voice commands, but these methods may have limitations in some cases. For example, for people with limited mobility or when their hands are busy, traditional control methods are not convenient enough. The emergence of brain-computer interface technology provides a new way to solve these problems.

[0003] Brain-computer interface technology is a technology that directly establishes communication between the brain and external devices. It detects the bioelectric signals generated by the brain and decodes these signals into machine-understandable instructions to achieve control of external devices. In smart home control systems, brain-computer interface technology allows users to control home appliances by thinking, such as turning on and off lights, adjusting temperature, controlling TVs, etc., greatly improving the convenience of life.

[0004] However, there are various noises in EEG signals. Existing denoising methods result in incomplete EEG signals, poor signal restoration, and poor accuracy of system judgment. Summary of the invention

[0005] Based on the technical problems existing in the background technology, the present invention proposes a smart home control system combined with a brain-computer interface.

[0006] The present invention proposes a smart home control system combined with a brain-computer interface, comprising an electroencephalogram signal acquisition device, a signal processing unit and a smart home control unit, wherein the electroencephalogram signal acquisition device is connected to the signal processing unit, the signal processing unit is connected to the smart home control unit, and the signal processing unit is also connected to a cloud server, wherein:

[0007] The EEG signal acquisition device is used to collect the user's EEG signals in real time, pre-process the EEG data using the fastest step descent IVA algorithm, and perform feature extraction through a common space pattern;

[0008] The signal processing unit is responsible for receiving and processing the digital signal transmitted by the EEG signal acquisition device, importing the data stream into the adaptive perceptual linear classifier to classify the imagination features and obtain the control instructions;

[0009] The smart home control unit is used to receive control instructions transmitted by the signal processing unit and control corresponding smart home devices according to the instructions.

[0010] Preferably, the EEG signal acquisition device is designed to be head-mounted or embedded, and captures electrical signals generated by brain activity through multiple electrodes. The collected EEG signals are EEG signals emitted by the scalp, and are converted into digital signals for transmission. The adaptive sample entropy wavelet threshold method is used to remove artifact components mixed in the EEG data, namely heartbeat, electromyography, and power frequency noise.

[0011] Preferably, the specific steps of the adaptive sample entropy wavelet threshold method include:

[0012] S11. Define a time series X = {x(1), x(2)…x(L)} with a length of L, and convert the time series X into a vector sequence X with a dimension of m. m (1), X m (2), …X m (L-m+1), where

[0013] X m (i)={x(i),x(i+1)…x(i+m-1)}

[0014] I=1,2…,L-m+1;

[0015] S12, when i≠j, vector X m (i) With vector X m The distance relationship of (j) is:

[0016]

[0017] S13, set the threshold value to r, D[X m (i),X m The ratio of the number of vectors with [j] < r to the total number of vectors is:

[0018]

[0019] After taking the average of the above formula, we get:

[0020]

[0021] S14. The sample entropy of the sequence is obtained as follows:

[0022]

[0023] S15. Construct an adaptive threshold function:

[0024]

[0025] Where λ is the threshold, d j,k is the detail coefficient before denoising, η(d j,k,λ,c) is the detail coefficient after denoising, c is an adjustable parameter, when c=1, it is a hard threshold function, when c=0, it is a soft threshold function;

[0026] S16, adaptively adjust the c value by using sample entropy, decompose the sequence of length L into Ln subsequences, calculate the sample entropy of each subsequence, and obtain the sequence {c 1 , c 2 , …c l}, bring it into the adaptive threshold function in step S15 to obtain the adaptive threshold function based on sample entropy:

[0027]

[0028] Preferably, the signal processing unit uses time-frequency analysis and power spectrum analysis methods to extract feature information related to the user's intention from the brain wave signal, and this feature information may include power and waveform characteristics of different frequency bands, and uses a machine learning model to perform intention recognition on the extracted feature information, wherein the machine learning model may be one of a neural network and a support vector machine. The machine learning model is used to train and learn the user's brain wave sample data for analysis and processing, and convert it into instructions that can be understood by the smart home system.

[0029] Preferably, the specific steps of extracting feature information include:

[0030] S21, collect the imagined movement EEG signals of a main trainer and m secondary trainers as training samples for training, wherein r training samples of each secondary trainer are selected from the EEG signals of the m secondary trainers, and the total number of training samples includes the main trainer's samples and m*r samples of all secondary trainers, and calculate the sum of the covariance matrix C of the main trainer's EEG signal j And the sum of the covariance matrices of the EEG signals of the trainees C j 'Respectively:

[0031]

[0032] When j is l(r), it represents the imagination of both hands, E ji represents the EEG signal of the master trainer's i-th imagination, E' ji represents the EEG signal of the trainee’s i-th imagination;

[0033] S22. Add the regularization parameters α and β to the sum of the covariance of the main trainer and the sum of the covariance of the secondary trainer to obtain:

[0034]

[0035] Where I is the N×N identity matrix, and N is the number of channels collected;

[0036] S23. Solve the regularized whitening matrix by summing the above mean-normalized covariance matrices and performing eigenvalue decomposition.

[0037]

[0038] R j (α,β) = P·R j (α,β)·P T = Uλ j U T

[0039] Obtain the projection matrix W = U T ·P, and obtain the features of the EEG signal:

[0040] Y j = W·E j .

[0041] Preferably, the steps for analyzing and processing the EEG sample data include:

[0042] S31. Set three initial values: Set the increment μ in the update criterion to a positive number, set the number of test samples x k to 0, and set the trained weight vector ω(k) to ω;

[0043] S32. Apply the newly input online test sample x k+1 to the adaptive linear classifier, and identify the value of x k , then calculate ω(k)·x k ;

[0044] S33. Adjust the value of the augmented weight vector ω(k + 1) according to the classification result:

[0045]

[0046] S34. If the step size k < n, change the value of the weight vector ω(k) to ω(k) = ω(k + 1), x k = x k+1 , and add the online test sample x k to the original training samples to form new sample data, and go back to step S32 to continue classification. When the step size k = n, the classification is completed.

[0047] Preferably, the data processed by the signal processing unit is stored in the cloud server.

[0048] Preferably, the smart home control unit communicates with multiple home appliances.

[0049] Preferably, the signal processing unit sends the generated smart home control instruction to the smart home control unit via a wireless communication protocol, wherein the wireless communication protocol is one of ZigBee and Wi-Fi.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. The EEG signal acquisition device collects the user's EEG signals in real time, and captures the electrical signals generated by brain activity through multiple electrodes, converts them into digital signals and transmits them to the signal processing unit to convert them into instructions that the smart home system can understand, and transmits the instructions to the smart home control unit to communicate with the smart home devices, so as to achieve the purpose of controlling home devices through thoughts. This interactive method is more natural and intuitive, without manual operation, and provides users with an unprecedented convenient experience;

[0052] 2. The use of brain-computer interface technology provides more accurate data support for the smart home system, enabling home appliances to more accurately understand and execute the user's intentions, thereby improving the intelligence level of the entire system. It is not only suitable for home environments, but can also be applied to rehabilitation medicine, game entertainment and other fields, providing users with more abundant and diverse services. For the elderly and disabled people with limited mobility, it enables them to control their homes more independently and improve their ability to take care of themselves;

[0053] The present invention uses brain-computer interface technology to analyze the user's brain activity in real time, understand the user's needs and intentions, and achieve the purpose of controlling home appliances through thoughts. This interactive method is more natural and intuitive, and does not require manual operation, providing users with an unprecedented convenient experience. For elderly people and disabled people with limited mobility, it enables them to control their homes more independently and improve their ability to take care of themselves. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A block diagram of a smart home control system combined with a brain-computer interface proposed by the present invention;

[0055] Figure 2 This is a flow chart of a smart home control system combined with a brain-computer interface proposed by the present invention. DETAILED DESCRIPTION

[0056] The present invention will be further explained below in conjunction with specific embodiments.

[0057] Example

[0058] Reference Figure 1-2This embodiment proposes a smart home control system combined with a brain-computer interface, including an EEG signal acquisition device, a signal processing unit and a smart home control unit. The EEG signal acquisition device is connected to the signal processing unit, wherein the EEG signal acquisition device is used to collect the user's EEG signals in real time. The device can be designed to be head-mounted or embedded to ensure the accuracy and stability of the signal. The EEG signal acquisition device captures the electrical signals generated by brain activity through multiple electrodes and converts them into digital signals for transmission; the original brain wave signals collected by the EEG signal acquisition device are amplified by an amplification circuit to increase the amplitude of the signal.

[0059] In this embodiment, an adaptive sample entropy wavelet threshold method is used to remove artifact components mixed in EEG data, namely heartbeat, electromyography, and power frequency noise. The specific steps include:

[0060] S11. Define a time series X = {x(1), x(2)…x(L)} with a length of L, and convert the time series X into a vector sequence X with a dimension of m. m (1), X m (2), …X m (L-m+1), where

[0061] X m (i)={x(i),x(i+1)…x(i+m-1)}

[0062] I=1,2…,L-m+1;

[0063] S12, when i≠j, vector X m (i) With vector X m The distance relationship of (j) is:

[0064]

[0065] S13, set the threshold value to r, D[X m (i),X m The ratio of the number of vectors with [j] < r to the total number of vectors is:

[0066]

[0067] After taking the average of the above formula, we get:

[0068]

[0069] S14. The sample entropy of the sequence is obtained as follows:

[0070]

[0071] S15. Construct an adaptive threshold function:

[0072]

[0073] Where λ is the threshold, d j,k is the detail coefficient before denoising, η(d j,k ,λ,c) is the detail coefficient after denoising, c is an adjustable parameter, when c=1, it is a hard threshold function, when c=0, it is a soft threshold function;

[0074] S16, adaptively adjust the c value by using sample entropy, decompose the sequence of length L into Ln subsequences, calculate the sample entropy of each subsequence, and obtain the sequence {c 1 , c 2 , …c l}, bring it into the adaptive threshold function in step S15 to obtain the adaptive threshold function based on sample entropy:

[0075]

[0076] The threshold function in this method should be a continuous function to avoid excessive additional oscillations; by introducing adaptive parameters, adaptive adjustments are made to signals with different noise levels, thereby achieving more flexible and better denoising; the detail coefficients with absolute values ​​larger than the threshold are as stable as possible, and the detail coefficients with absolute values ​​smaller than the threshold are minimized as much as possible, thereby achieving better denoising effects while retaining the complete signal. The signal obtained by the adaptive sample entropy wavelet threshold denoising method can effectively solve the problem of large additional oscillations in hard threshold denoising, and the obtained signal curve is smoother. It can also solve the defects of "burrs" and excessive peak reduction in soft threshold denoising, making the signal restoration better.

[0077] The signal processing unit is responsible for receiving and processing the digital signals transmitted by the EEG signal acquisition device. The signal processing unit identifies the user's intentions through specific algorithms and models, and converts them into instructions that can be understood by the smart home system. These algorithms and models can be customized and optimized according to different user needs and application scenarios. The signal processing unit uses time-frequency analysis, power spectrum analysis and other methods to extract feature information related to the user's intentions from the EEG signal. This feature information may include power and waveform characteristics in different frequency bands, etc., and uses a machine learning model to perform intent recognition on the extracted feature information, where the machine learning model can be one of a neural network and a support vector machine. The machine learning model is used to train and learn the mapping relationship between the user's brain wave patterns and home control instructions.

[0078] The specific steps of extracting feature information include:

[0079] S21, collect the imagined movement EEG signals of a main trainer and m secondary trainers as training samples for training, wherein r training samples of each secondary trainer are selected from the EEG signals of the m secondary trainers, and the total number of training samples includes the main trainer's samples and m*r samples of all secondary trainers, and calculate the sum of the covariance matrix C of the main trainer's EEG signal j And the sum of the covariance matrices of the EEG signals of the trainees C j 'Respectively:

[0080]

[0081] When j is l(r), it represents the imagination of both hands, E ji represents the EEG signal of the master trainer's i-th imagination, E' ji represents the EEG signal of the trainee’s i-th imagination;

[0082] S22. Add the regularization parameters α and β to the sum of the covariance of the main trainer and the sum of the covariance of the secondary trainer to obtain:

[0083]

[0084] Where I is the N×N identity matrix, and N is the number of channels collected;

[0085] S23, solving the regularized whitening matrix by summing the above mean-normalized covariance matrix and performing eigenvalue decomposition,

[0086]

[0087] R j (α,β)=P·R j (α,β)·P T =Uλ j U T

[0088] Get the projection matrix W = U T P, and obtain the characteristics of the EEG signal:

[0089] Y j =W·E j .

[0090] The data of other sub-trainers are added to the data of the main trainer in a reasonable way, so that the EEG signal data of the main trainer is no longer a single sample, which increases the diversity of the experimental data and correspondingly improves the stability of the system.

[0091] In this embodiment, the steps of analyzing and processing the brain wave sample data include:

[0092] S31. Set three initial values: Set the increment μ in the update criterion to a positive number, set the number of test samples x k to 0, and set the trained weight vector ω(k) to ω;

[0093] S32. Apply the newly input online test sample x k+1 to the adaptive linear classifier and identify the value of x k , then calculate ω(k)·x k ;

[0094] S33. Adjust the value of the augmented weight vector ω(k + 1) according to the classification result:

[0095]

[0096] S34. If the step k < n, change the value of the weight vector ω(k) to ω(k) = ω(k + 1), x k = x k+1 , and add the online test sample x k to the original training samples to form new sample data, and go back to step S32 to continue classification. When the step k = n, the classification is completed.

[0097] The adaptive perceptron linear classifier recombines the latest collected sample data into the original sample data through calculation, enabling the sample data collected in the online experiment to adapt to the current system in a more optimized form, thereby improving its classification accuracy.

[0098] The signal processing unit is connected to the smart home control unit. The signal processing unit sends the generated smart home control instructions to the smart home control unit through a wireless communication protocol, where the wireless communication protocol is one of ZigBee and Wi-Fi. The smart home control unit is used to receive the instructions transmitted by the signal processing unit and control the corresponding smart home devices according to the instructions. The smart home control unit can communicate with a variety of home devices, such as lighting systems, air conditioning systems, curtain systems, security monitoring systems, etc. Users can directly control the on / off, adjustment, etc. of these devices through brain signals;

[0099] The signal processing unit is also connected to a cloud server, and the data processed by the signal processing unit can be stored in the cloud server;

[0100] This embodiment adopts brain-computer interface technology to be able to analyze the user's brain activities in real time, understand the user's needs and intentions, and achieve the purpose of controlling home devices through thoughts. This interaction method is more natural and intuitive, without the need for manual operation, providing users with an unprecedented convenient experience. For the elderly and disabled with limited mobility, it enables them to control their houses more independently and improve their self-care ability.

[0101] In this embodiment, when in use, the user first wears the EEG signal acquisition device (such as an EEG cap) to ensure that the EEG signal acquisition device is worn correctly and comfortably, and then the user starts the smart home control system, the smart home control system starts the self-check and initialization process, the smart home control system checks the connection status of the EEG signal acquisition device to ensure that the device has been successfully connected and is in working condition, the EEG signal acquisition device starts to collect the user's brain wave signal in real time, the collected raw brain wave signal is amplified by the amplifier circuit, and the signal is filtered by the filter circuit to remove noise and interference, the filtered signal is converted into a digital signal by an analog-to-digital converter (ADC), and the digital signal is sent to the signal processing unit for further processing and analysis;

[0102] The signal processing unit uses time-frequency analysis, power spectrum analysis and other methods to extract feature information related to the user's intention from the processed and analyzed brain wave signals. These feature information may include power in different frequency bands, waveform characteristics, etc. The signal processing unit uses a machine learning model to identify the extracted feature information. According to the identified user intention, the corresponding smart home control instruction is generated and transmitted to the smart home control unit. The smart home control unit sends it to the corresponding smart home device through the wireless communication protocol according to the instruction. After receiving the instruction, the smart home device performs the corresponding operation, such as turning on and off the light, adjusting the temperature, etc. After the smart home device completes the execution, the execution status is fed back to the smart home control system through the wireless communication network, thereby achieving the purpose of controlling home appliances through thoughts. This interactive method is more natural and intuitive, without manual operation, providing users with an unprecedented convenient experience. By analyzing the user's brain activity in real time, the user's needs and intentions are understood, thereby providing users with more personalized services. For the elderly and disabled people with limited mobility, they can control their houses more independently and improve their ability to take care of themselves.

[0103] Through the specific application examples of this embodiment:

[0104] 1. Control light brightness, color, etc.

[0105] Brain-computer interfaces can control the brightness and color of lights through thinking, providing users with a more convenient and personalized lighting experience. For example, when a user wants to brighten the lights, the brain-computer interface device detects a specific thinking pattern in the user's brain, converts it into a control signal and sends it to the lighting system, thereby increasing the brightness of the lights. In the process of achieving this, non-invasive brain-computer interfaces usually obtain the user's EEG signals through scalp EEG acquisition devices, which reflect the user's brain activity state. The signal processing module analyzes the collected EEG signals and extracts features related to light control. For example, brain waves of a specific frequency may represent the user's intention to increase the brightness of the light. Then, these features are converted into specific control instructions, such as commands to increase the brightness of the light, through a pre-set algorithm. The control of the color of the light can also be achieved through the user's thinking. When the user imagines a specific color, the brain-computer interface device captures the corresponding EEG signal changes and converts them into instructions to control the color of the light. For example, according to some research and practice, different colors may be associated with specific brain wave patterns, and the brain-computer interface system can control the color of the light by identifying these patterns. In addition, some advanced brain-computer interface technologies can also automatically adjust the brightness and color of the light in combination with the user's mood and environmental factors. For example, when the user is in a relaxed state, the light may automatically dim and change to a warm tone; when the user needs to concentrate, the light may become brighter and adjusted to a cool tone.

[0106] 2. Control the air conditioning temperature, etc.

[0107] The way brain-computer interface controls the temperature of air conditioners is mainly to determine the user's temperature needs by detecting the user's brain signals and converting them into instructions to control the air conditioner. For example, when the user feels hot, the brain will generate specific EEG signals, which the brain-computer interface device can capture and interpret as instructions to increase the temperature of the air conditioner. Non-invasive brain-computer interfaces usually use scalp EEG acquisition devices to obtain the user's EEG signals. These devices can detect the weak electrical signals generated by brain activity and transmit them to the signal processing module for analysis. During the analysis process, the system looks for EEG features related to temperature control. For example, brain waves of a specific frequency may indicate the user's feeling of temperature, such as hot or cold. Once the user's temperature needs are determined, the brain-computer interface system will convert them into instructions to control the air conditioner. These instructions can be sent to the air conditioning system via wireless communication to achieve temperature adjustment. In addition, some brain-computer interface systems can also automatically adjust the air conditioning temperature based on environmental factors and user habits. For example, the system can automatically adjust the temperature of the air conditioner based on environmental parameters such as indoor and outdoor temperature and humidity and the user's previous temperature setting habits to provide a more comfortable environment.

[0108] 3. Control TV channels, etc.

[0109] The principle of brain-computer interface controlling TV channels is to determine the user's intention to select TV channels by detecting the user's brain signals and converting them into instructions to control the TV. When the user wants to switch TV channels, the brain will generate specific EEG signals, which can be captured by the brain-computer interface device. Non-invasive brain-computer interfaces usually use scalp EEG acquisition devices to obtain the user's EEG signals. These devices can detect the weak electrical signals generated by brain activity and transmit them to the signal processing module for analysis. During the analysis process, the system looks for EEG features related to TV channel control. For example, a specific brain wave pattern may represent that the user wants to switch to a specific TV channel. Once the user's channel selection intention is determined, the brain-computer interface system will convert it into instructions to control the TV. These instructions can be sent to the TV via wireless communication to achieve channel switching. In addition, some brain-computer interface systems can also automatically recommend TV channels based on the user's viewing history and interest preferences. For example, the system can automatically recommend TV channels that may be of interest to the user based on information such as the channel types and program preferences that the user has watched in the past, thereby improving the user's viewing experience.

[0110] 4. Send an alarm message.

[0111] When the owner is not at home, the brain-computer interface can send alarm information in a variety of ways. For example, when the smart home security monitoring system detects an abnormal situation, the brain-computer interface device can collect the user's EEG signals and convert them into alarm information and send them to the owner. Specifically, the technology mentioned in the new brain-computer interface patent announced by Huawei can be used to quickly identify the user's intentions through real-time monitoring and analysis of brain signals. When the system detects an abnormal situation, the brain-computer interface device will capture the specific thinking patterns in the user's brain and convert them into control signals to send to the owner's mobile devices, such as mobile phones, tablets, etc., to alert the owner of abnormal conditions at home.

[0112] In addition, we can also learn from the technology in the brain-computer interface alarm method and system, and induce EEG signals by generating visual stimulation codes to improve reaction speed and reduce false alarm rate. When the security monitoring system detects an abnormal situation, the visual stimulation module generates a specific visual stimulation code, the acquisition module collects the EEG signals generated by the user in response to the visual stimulation code, and the EEG analysis module continuously compares the currently collected EEG signals with the preset EEG template signals. Once the proportion of the results matching the visual stimulation code in the comparison results reaches the set value, the corresponding alarm signal is output and sent to the owner.

[0113] 5. Link security systems for defense.

[0114] When an abnormal situation occurs, the brain-computer interface can be linked to the security system for defense. For example, when the security monitoring system detects that someone has broken into the house, the brain-computer interface can automatically identify and alarm, and at the same time link the security system for defense. Specifically, the technology of the automatic indoor light adjustment device based on the brain-computer interface can be used to collect the current EEG signal generated by the user under the current light stimulation through the EEG acquisition module. The EEG processing module generates a control instruction based on the corresponding relationship between the current EEG signal and the EEG signal in the database and the indoor light brightness. The control module adjusts the indoor light brightness according to the control instruction to interfere with the intruder's line of sight.

[0115] At the same time, it can also link other security devices in the smart home, such as door and window sensors, cameras, alarms, etc. When the brain-computer interface detects an abnormal situation, the control module can control the door and window sensors according to the control instructions, automatically close the doors and windows, and prevent the intruder from further entering the home. The camera can automatically turn on, record the intruder's behavior, and send the video information to the owner's mobile device. The alarm can sound an alarm to deter the intruder and alert the surrounding neighbors and property staff.

[0116] In addition, we can also learn from the technology of the "Brain-Computer AI Smart Ward" of the Pazhou Laboratory, collect multimodal signals such as electrooculogram signals, electroencephalogram signals, and head movement signals through a multimodal brain-computer headband, and analyze these signals by computer fusion to achieve complex environmental control. When an abnormal situation occurs, the brain-computer interface can automatically identify the type and location of the abnormal situation through the analysis of multimodal signals, and link the security system for targeted defense. For example, if a fire hazard is detected, the brain-computer interface can automatically trigger the alarm device, notify the user and take appropriate emergency measures, such as automatically opening windows for ventilation and closing gas valves.

[0117] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A smart home control system combined with a brain-computer interface, characterized in that: It includes an EEG signal acquisition device, a signal processing unit and a smart home control unit. The EEG signal acquisition device is connected to the signal processing unit, the signal processing unit is connected to the smart home control unit, and the signal processing unit is also connected to a cloud server, wherein: The EEG signal acquisition device is used to collect the user's EEG signals in real time, pre-process the EEG data using the fastest step descent IVA algorithm, and perform feature extraction through the common space mode; The signal processing unit is responsible for receiving and processing the digital signal transmitted by the EEG signal acquisition device, importing the data stream into the adaptive perceptual linear classifier to classify the imagination features and obtain the control instructions; The smart home control unit is used to receive control instructions transmitted by the signal processing unit and control corresponding smart home devices according to the instructions.

2. According to claim 1, a smart home control system combined with a brain-computer interface is characterized in that: The EEG signal acquisition device is designed to be head-mounted or embedded, and captures electrical signals generated by brain activity through multiple electrodes. The collected EEG signals are EEG signals emitted by the scalp, and are converted into digital signals for transmission. The adaptive sample entropy wavelet threshold method is used to remove artifact components mixed in the EEG data, namely heartbeat, electromyography, and power frequency noise.

3. According to claim 2, a smart home control system combined with a brain-computer interface is characterized in that: The specific steps of the adaptive sample entropy wavelet threshold method include: S11. Define a time series X = {x(1), x(2)…x(L)} with a length of L, and convert the time series X into a vector sequence X with a dimension of m. m (1), X m (2), …X m (L-m+1), where X m (i)={x(i),x(i+1)…x(i+m-1)} I=1,2…,L-m+1; S12, when i≠j, vector X m (i) With vector X m The distance relationship of (j) is: S13, set the threshold value to r, D[X m (i),X m The ratio of the number of vectors with [j] < r to the total number of vectors is: After taking the average of the above formula, we get: S14. The sample entropy of the sequence is obtained as follows: S15. Construct an adaptive threshold function: Where λ is the threshold, d j,k is the detail coefficient before denoising, η(d j,k ,λ,c) is the detail coefficient after denoising, c is an adjustable parameter, when c=1, it is a hard threshold function, when c=0, it is a soft threshold function; S16, adaptively adjust the c value by using sample entropy, decompose the sequence of length L into Ln subsequences, calculate the sample entropy of each subsequence, and obtain the sequence {c1, c2, ...c l }, bring it into the adaptive threshold function in step S15 to obtain the adaptive threshold function based on sample entropy:

4. According to claim 1, a smart home control system combined with a brain-computer interface is characterized in that: The signal processing unit uses time-frequency analysis and power spectrum analysis methods to extract feature information related to the user's intention from the brain wave signal. The feature information may include power and waveform characteristics in different frequency bands, and uses a machine learning model to identify the intention of the extracted feature information. The machine learning model can be one of a neural network and a support vector machine. The machine learning model is used to train and learn the user's brain wave sample data for analysis and processing, and convert it into instructions that can be understood by the smart home system.

5. The smart home control system combined with a brain-computer interface according to claim 4, characterized in that: The specific steps of extracting feature information include: S21, collect the imagined movement EEG signals of a main trainer and m secondary trainers as training samples for training, wherein r training samples of each secondary trainer are selected from the EEG signals of the m secondary trainers, and the total number of training samples includes the main trainer's samples and m*r samples of all secondary trainers, and calculate the sum of the covariance matrix C of the main trainer's EEG signal j And the sum of the covariance matrices of the EEG signals of the trainees C j 'Respectively: When j is l(r), it represents the imagination of both hands, E ji represents the EEG signal of the master trainer's i-th imagination, E' ji represents the EEG signal of the trainee’s i-th imagination; S22. Add the regularization parameters α and β to the sum of the covariance of the main trainer and the sum of the covariance of the secondary trainer to obtain: Where I is the N×N identity matrix, and N is the number of channels collected; S23, solving the regularized whitening matrix by summing the covariance matrix in step S22 and performing eigenvalue decomposition, R j (a, b)=P·R j (a, b)·P T =Uλ j U T Get the projection matrix W = U T P, and obtain the characteristics of the EEG signal: Y j =W E j 。 6. The smart home control system combined with a brain-computer interface according to claim 4, characterized in that: The steps for analyzing and processing brain wave sample data include: S31, set three initial values: set the increment μ in the update criterion to a positive number, test sample x k The number is set to 0, and the training weight vector ω(k) is set to ω; S32, the newly input online test sample x k+1 Applied to the adaptive linear classifier and identifies x k Then calculate ω(k)·x k ; S33. Adjust the value of the augmented weight vector ω(k+1) according to the classification result: S34. If the step size k < n, change the value of the weight vector ω(k) to ω(k) = ω(k + 1), x k = x k+1 , and add the online test sample x k to the original training samples to form new sample data, and go back to step S32 to continue classification. When the step size k = n, the classification is completed.

7. The smart home control system combined with a brain-computer interface according to claim 1, characterized in that: The data processed by the signal processing unit is stored in the cloud server.

8. The smart home control system combined with a brain-computer interface according to claim 1, characterized in that: The smart home control unit communicates with a variety of home devices.

9. The smart home control system combined with a brain-computer interface according to claim 1, characterized in that: The signal processing unit sends the generated smart home control instruction to the smart home control unit via a wireless communication protocol, wherein the wireless communication protocol is one of ZigBee and Wi-Fi.

Images