Multi-device smart home intention priority decision-making method based on electroencephalogram signals

Abstract

The invention discloses a multi-device smart home intention priority decision-making method based on electroencephalogram signals, and the method comprises the following steps: S1, signal purification: carrying out the purification of original electroencephalogram signals through a signal purification layer, and extracting effective electroencephalogram features; s2, intention analysis: converting the electroencephalogram features extracted in the S1 into an understandable equipment control instruction through an intention analysis layer; and S3, intelligent decision making: an intelligent decision making layer adopts a three-in-one framework of pre-judgment, execution and error correction. According to the method, a three-stage processing mechanism of signal purification, intention analysis and dynamic decision is adopted, the multi-intention conflict rate can be effectively reduced, a self-adaptive wavelet noise reduction technology is adopted during signal purification, the signal-to-noise ratio is remarkably improved, the recognition error rate of motion intention characteristic waves is reduced, and in addition, by introducing a dynamic weight matrix, the recognition accuracy of the motion intention characteristic waves is improved. Dynamic balance among safety, energy conservation and habits is achieved, and decision reasonability and user satisfaction are improved.

Description

Technical Field

[0001] This invention relates to the field of smart home technology, and in particular to a multi-device smart home intent priority decision-making method based on electroencephalogram (EEG) signals. Background Technology

[0002] With the development of intelligence, the technology of controlling smart home devices based on brain-computer interface (BCI) has gradually emerged. Using brain-computer interface (BCI) technology, the smart home devices are controlled by detecting and interpreting the user's brainwave (EEG) signals. The existing brain-computer interface control smart home system has the following shortcomings in use: 1. Single signal defect: The existing brain-computer interface control smart home system directly maps to a single device. When the user generates multiple control intentions at the same time, the system will cause device conflicts due to signal aliasing; 2. Environmental interference problem: The traditional method uses fixed frequency band filtering (such as 8-30Hz), but the electromagnetic noise of home appliances and the delta wave (0.5-4Hz) cause frequency band overlap, resulting in a recognition error rate of up to 32% for motion intention feature wave (μ wave); (3) Lack of decision-making mechanism: There is a lack of inter-device collaborative logic. For example, when the user intends to "turn on the air purifier", the system will not automatically close the window and the linkage rules need to be set manually. In view of the above problems, this application proposes a multi-device smart home intention priority decision method based on brainwave signals. Summary of the Invention

[0003] Based on the technical problems existing in the background technology, the present invention proposes a multi-device smart home intention priority decision-making method based on electroencephalogram (EEG) signals.

[0004] This invention proposes a multi-device smart home intent prioritization decision-making method based on electroencephalogram (EEG) signals, comprising the following steps:

[0005] S1: Signal purification: The original EEG signal is purified through the signal purification layer, and effective EEG features are extracted;

[0006] S2: Intent parsing: The intent parsing layer converts the EEG features extracted in S1 into understandable device control commands.

[0007] S3: Intelligent Decision-Making: The intelligent decision-making layer adopts a three-in-one architecture of "prediction-execution-error correction" to transform the understandable device control commands in S2 into precise and efficient control commands;

[0008] S4: Perform equipment actions: Perform operations on the equipment according to control instructions.

[0009] Preferably, the specific steps of S1 are as follows:

[0010] S101: Feature Wave Extraction: Focus on identifying two types of brain waves directly related to motor intentions, including μ waves (8-12Hz) and β waves (18-26Hz). μ waves are the "motor preparation waves" that appear when imagining limb movements, and β waves are the "precision control waves" when concentrating attention.

[0011] S102: Noise Suppression: The adaptive notch filter is designed to eliminate power frequency interference. The formula used is:

[0012]

[0013] To address 50Hz power frequency interference, the center frequency f0 = 50Hz and the damping coefficient r = 0.99 are set. This achieves 20dB interference suppression while preserving the characteristic wave, specifically eliminating 50Hz electromagnetic interference from fluorescent lights and household appliances, while retaining clear brainwave signals.

[0014] S103: Focusing on the μ-wave and β-wave features in S101, wavelet packet decomposition is used to extract motion-related rhythms.

[0015] Preferably, the specific steps of S2 are as follows:

[0016] S201: Device Energy Field Localization: Each device is abstracted as an energy node, the propagation attenuation equation of EEG intention signal is defined, and the intensity of user attention to each device is calculated through the equation formula;

[0017] S202: Dynamic priority decision matrix, introducing an "intelligent scoring card" mechanism to calculate the priority ranking of control commands for each device in real time and refresh the device execution order every second;

[0018] S203: The execution logic is set according to priority. The execution logic is as follows: the system refreshes the score every 200ms. When both the gas shut-off command (68 points) and the air conditioning turn-on command (50 points) are detected at the same time, the gas shut-off operation with the higher safety score is executed first.

[0019] Preferably, in step S201, the propagation attenuation equation for the EEG intention signal is as follows: By determining the device a user truly wants to control based on spatial distance and brainwave intensity, the optimal balance can be found between safety, energy efficiency, and user habits.

[0020] Where P EEG The power of the EEG signal is expressed in μV. 2 / Hz, G i The weights are for different equipment types, with security equipment having a weight of 1.5 and environmental equipment having a weight of 1.0. i The spatial distance between the user and the device is obtained through UWB positioning, and α is the attenuation factor with a value of 2.5.

[0021] Preferably, in step S202, the "intelligent scoring card" mechanism is shown in the table below:

[0022]

[0023]

[0024] Preferably, in step S202, the formula used to calculate the priority order of each device control command is:

[0025] Where S i For safety factor, E i H is the energy consumption coefficient. i λ is the habit coefficient, i.e., the historical usage frequency, and λ is the dynamic weight, ∑λ=1∑λ=1, which is adjusted every hour according to the environmental conditions;

[0026] The dynamic weight adjustment rule can be obtained from the above formula:

[0027] Basic weighting: Safety 40% + Energy saving 30% + Habits 30%;

[0028] During a fire alarm: 60% safety + 20% energy saving + 20% habitual behavior;

[0029] In power-saving mode: Safety 30% + Energy Saving 50% + Habitual 20%

[0030] Preferably, the detailed working mechanism of each module in the "prediction-execution-error correction" three-in-one architecture in S3 is as follows:

[0031] The prediction engine uses LSTM (Long Short-Term Memory) network to predict the user's intention trend in the next 3 seconds and preloads device drivers in advance, similar to how mobile phone input methods predict the next word, reducing waiting time.

[0032] The execution engine adopts a hierarchical instruction distribution mechanism control strategy in combination with a dynamic adjustment mechanism. Its hierarchical instructions include high-priority instructions, medium-priority instructions and low-priority instructions. High-priority instructions include security equipment and environmental equipment, medium-priority instructions include energy-consuming equipment, and low-priority instructions include entertainment equipment. Its dynamic adjustment mechanism includes (1) automatically merging similar instructions in power-saving mode; (2) immediately suspending non-essential equipment operation when an accidental fall is detected.

[0033] Error correction engine: When the deviation between the actual EEG signal and the predicted value is greater than 15%, the system automatically adjusts the control parameters. The formula used is:

[0034] Where L is the intention recognition loss function, calculated using cross-entropy, and η is the learning rate, with an initial value of 0.01 that decays with the number of iterations.

[0035] Preferably, the error correction process of the error correction engine is as follows:

[0036] S301: Deviation detection: Compare the actual status of the equipment with the expected value of the instruction every 50ms;

[0037] S302: Root cause analysis: If the signal transmission delay is greater than the threshold, activate local caching to accelerate; if the device is unresponsive, switch to the backup control channel; if the user's intent changes, recalculate the priority.

[0038] S303: Parameter Adjustment: Perform short-term adjustments and long-term optimizations. Short-term adjustments modify the current control command parameters; long-term optimizations update the prediction model weights.

[0039] Compared with existing technologies, the beneficial effects of this invention are:

[0040] 1. By using a dynamic energy field sorting mechanism, the conflict rate is effectively reduced, the user experience is improved, and the problem of device conflict caused by multiple intentions in existing technologies is solved.

[0041] 2. By using adaptive wavelet denoising technology, the signal-to-noise ratio is significantly improved, the error rate of motion intention feature wave recognition is reduced, and the problem of existing technologies using fixed frequency band filtering being susceptible to electromagnetic noise interference from household appliances is solved.

[0042] 3. By introducing a dynamic weight matrix, a dynamic balance between security, energy saving, and user habits is achieved, which improves the rationality of decision-making and user satisfaction. Furthermore, through the prediction engine and hierarchical instruction distribution mechanism, rapid response and execution of instructions are achieved.

[0043] This invention employs a three-level processing mechanism of "signal purification → intent parsing → dynamic decision-making," which effectively reduces the multi-intent conflict rate. Furthermore, the use of adaptive wavelet denoising technology during signal purification significantly improves the signal-to-noise ratio and reduces the error rate in identifying motion intent feature waves. In addition, by introducing a dynamic weight matrix, a dynamic balance is achieved between security, energy saving, and user habits, thereby enhancing the rationality of decision-making and user satisfaction. Attached Figure Description

[0044] Figure 1 This is a flowchart of a multi-device smart home intent prioritization decision-making method based on electroencephalogram (EEG) signals proposed in this invention;

[0045] Figure 2 This is a flowchart illustrating the intent parsing process in a multi-device smart home intent priority decision-making method based on electroencephalogram (EEG) signals proposed in this invention.

[0046] Figure 3 This is a flowchart of the three-engine collaborative workflow in a multi-device smart home intent priority decision-making method based on electroencephalogram (EEG) signals proposed in this invention.

[0047] Figure 4 This is a flowchart illustrating the technical features of a multi-device smart home intent priority decision-making method based on electroencephalogram (EEG) signals proposed in this invention. Detailed Implementation

[0048] The present invention will be further explained below with reference to specific embodiments.

[0049] Example

[0050] Reference Figure 1-4 This embodiment proposes a multi-device smart home intent prioritization decision-making method based on electroencephalogram (EEG) signals, including the following steps:

[0051] S1: Signal purification: The original EEG signal is purified through the signal purification layer, and effective EEG features are extracted;

[0052] The specific steps are as follows:

[0053] S101: Feature Wave Extraction: Focus on identifying two types of brain waves directly related to motor intentions, including μ waves (8-12Hz) and β waves (18-26Hz). μ waves are the "motor preparation waves" that appear when imagining limb movements, and β waves are the "precision control waves" when concentrating attention.

[0054] S102: Noise Suppression: The adaptive notch filter is designed to eliminate power frequency interference. The formula used is:

[0055]

[0056] To address 50Hz power frequency interference, the center frequency f0 = 50Hz and the damping coefficient r = 0.99 are set. This achieves 20dB interference suppression while preserving the characteristic wave, specifically eliminating 50Hz electromagnetic interference from fluorescent lights and household appliances, while retaining clear brainwave signals.

[0057] S103: Focusing on the μ-wave and β-wave features in S101, wavelet packet decomposition is used to extract motion-imagination-related rhythms;

[0058] S2: Intent parsing: The intent parsing layer converts the EEG features extracted in S1 into understandable device control commands.

[0059] The specific steps are as follows:

[0060] S201: Device Energy Field Localization: Each device is abstracted as an energy node, the propagation attenuation equation of EEG intention signal is defined, and the intensity of user attention to each device is calculated through the equation formula;

[0061] The propagation attenuation equation for the EEG intention signal is as follows: By determining the device a user truly wants to control based on spatial distance and brainwave intensity, the optimal balance can be found between safety, energy efficiency, and user habits.

[0062] Where P EEG The power of the EEG signal is expressed in μV. 2 / Hz, G i The weights are for different equipment types, with security equipment having a weight of 1.5 and environmental equipment having a weight of 1.0. i The spatial distance between the user and the device is obtained through UWB positioning, and α is the attenuation factor with a value of 2.5.

[0063] S202: Dynamic priority decision matrix, introducing an "intelligent scoring card" mechanism to calculate the priority ranking of control commands for each device in real time and refresh the device execution order every second;

[0064] Its "intelligent scoring card" mechanism is shown in the table below:

[0065]

[0066] The formula used to calculate the priority order of control commands for each device is:

[0067] Where S i For safety factor, E i H is the energy consumption coefficient. i λ is the habit coefficient, i.e., the historical usage frequency, and λ is the dynamic weight, ∑λ=1∑λ=1, which is adjusted every hour according to the environmental conditions;

[0068] The dynamic weight adjustment rule can be obtained from the above formula:

[0069] Basic weighting: Safety 40% + Energy saving 30% + Habits 30%;

[0070] During a fire alarm: 60% safety + 20% energy saving + 20% habitual behavior;

[0071] In power-saving mode: Safety 30% + Energy Saving 50% + Habitual 20%;

[0072] S203: The execution logic is set according to priority. The execution logic is as follows: the system refreshes the score every 200ms. When the gas shut-off command (68 points) and the air conditioner turn-on command (50 points) are detected at the same time, the gas shut-off operation with the higher safety score is executed first.

[0073] S3: Intelligent Decision-Making: The intelligent decision-making layer adopts a three-in-one architecture of "prediction-execution-error correction" to transform the understandable device control commands in S2 into precise and efficient control commands;

[0074] The detailed working mechanism of each module in the "prediction-execution-error correction" three-in-one architecture is as follows:

[0075] The prediction engine uses LSTM (Long Short-Term Memory) network to predict the user's intention trend in the next 3 seconds and preloads device drivers in advance, similar to how mobile phone input methods predict the next word, reducing waiting time.

[0076] The execution engine adopts a hierarchical instruction distribution mechanism control strategy in combination with a dynamic adjustment mechanism. Its hierarchical instructions include high-priority instructions, medium-priority instructions and low-priority instructions. High-priority instructions include security equipment and environmental equipment, medium-priority instructions include energy-consuming equipment, and low-priority instructions include entertainment equipment. Its dynamic adjustment mechanism includes (1) automatically merging similar instructions in power-saving mode; (2) immediately suspending non-essential equipment operation when an accidental fall is detected.

[0077] Error correction engine: When the actual EEG signal deviates from the predicted value by more than 15% (such as when the curtains are not closed on time), the system automatically adjusts the control parameters, similar to the real-time trajectory correction of an autonomous vehicle. The formula used is:

[0078] Where L is the intention recognition loss function, calculated using cross-entropy, and η is the learning rate, with an initial value of 0.01 that decays with the number of iterations;

[0079] The error correction process of its error correction engine is as follows:

[0080] S301: Deviation detection: Compare the actual status of the equipment with the expected value of the instruction every 50ms;

[0081] S302: Root cause analysis: If the signal transmission delay is greater than the threshold, activate local caching to accelerate; if the device is unresponsive, switch to the backup control channel; if the user's intent changes, recalculate the priority.

[0082] S303: Parameter Adjustment: Performs short-term adjustments and long-term optimizations. Short-term adjustments modify the current control command parameters; long-term optimizations update the prediction model weights.

[0083] S4: Perform equipment actions: Perform operations on the equipment according to control commands;

[0084] This embodiment effectively reduces the multi-intent conflict rate by adopting a three-level processing mechanism of "signal purification → intent parsing → dynamic decision-making". In addition, the adaptive wavelet denoising technology is used in signal purification, which significantly improves the signal-to-noise ratio and reduces the error rate of motion intent feature wave recognition. Furthermore, by introducing a dynamic weight matrix, a dynamic balance between security, energy saving, and user habits is achieved, which improves the rationality of decision-making and user satisfaction.

[0085] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-device smart home intent prioritization decision-making method based on electroencephalogram (EEG) signals, characterized in that, Includes the following steps: S1: Signal purification: The original EEG signal is purified through the signal purification layer, and effective EEG features are extracted; S2: Intent parsing: The intent parsing layer converts the EEG features extracted in S1 into understandable device control commands. S3: Intelligent Decision-Making: The intelligent decision-making layer adopts a three-in-one architecture of "prediction-execution-error correction" to transform the understandable device control commands in S2 into precise and efficient control commands; S4: Perform equipment actions: Perform operations on the equipment according to control instructions.

2. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Feature Wave Extraction: Focus on identifying two types of brain waves directly related to motor intentions, including μ waves and β waves. μ waves are the "motor preparation waves" that appear when imagining limb movements, and β waves are the "precision control waves" when concentrating attention. S102: Noise Suppression: The adaptive notch filter is designed to eliminate power frequency interference. The formula used is: To address 50Hz power frequency interference, the center frequency f0 = 50Hz and the damping coefficient r = 0.99 are set. This achieves 20dB interference suppression while preserving the characteristic wave, specifically eliminating 50Hz electromagnetic interference from fluorescent lights and household appliances, while retaining clear brainwave signals. S103: Focusing on the μ-wave and β-wave features in S101, wavelet packet decomposition is used to extract motion-related rhythms.

3. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Device Energy Field Localization: Each device is abstracted as an energy node, the propagation attenuation equation of EEG intention signal is defined, and the intensity of user attention to each device is calculated through the equation formula; S202: Dynamic priority decision matrix, introducing an "intelligent scoring card" mechanism to calculate the priority ranking of control commands for each device in real time and refresh the device execution order every second; S203: The execution logic is set according to priority. The execution logic is as follows: the system refreshes the score every 200ms. When both the gas shut-off command and the air conditioning turn-on command are detected at the same time, the gas shut-off operation with the higher safety score is executed first.

4. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 3, characterized in that, In S201, the propagation attenuation equation for the EEG intention signal is as follows: By determining the device a user truly wants to control based on spatial distance and brainwave intensity, the optimal balance can be found between safety, energy efficiency, and user habits. Where P EEG The power of the EEG signal is expressed in μV. 2 / Hz, G i The weights are for different equipment types, with security equipment having a weight of 1.5 and environmental equipment having a weight of 1.

0. i The spatial distance between the user and the device is obtained through UWB positioning, and α is the attenuation factor with a value of 2.

5.

5. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 3, characterized in that, In step S202, the "intelligent scoring card" mechanism is shown in the table below:

6. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 3, characterized in that, In step S202, the formula used to calculate the priority order of each device control command is as follows: Where S i For safety factor, E i H is the energy consumption coefficient. i λ is the habit coefficient, i.e., the historical usage frequency, and λ is the dynamic weight, ∑λ=1∑λ=1, which is adjusted every hour according to the environmental conditions; The dynamic weight adjustment rule can be obtained from the above formula: Basic weighting: Safety 40% + Energy saving 30% + Habits 30%; During a fire alarm: 60% safety + 20% energy saving + 20% habitual behavior; In power-saving mode: Safety 30% + Energy Saving 50% + Habitual 20% 7. The method for prioritizing intentions in multi-device smart home systems based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The detailed working mechanism of each module in the "prediction-execution-error correction" three-in-one architecture of S3 is as follows: The prediction engine uses LSTM (Long Short-Term Memory) network to predict the user's intention trend in the next 3 seconds and preloads device drivers in advance, similar to how mobile phone input methods predict the next word, reducing waiting time. The execution engine adopts a hierarchical instruction distribution mechanism control strategy in combination with a dynamic adjustment mechanism. Its hierarchical instructions include high-priority instructions, medium-priority instructions and low-priority instructions. High-priority instructions include security equipment and environmental equipment, medium-priority instructions include energy-consuming equipment, and low-priority instructions include entertainment equipment. Its dynamic adjustment mechanism includes (1) automatically merging similar instructions in power-saving mode; (2) immediately suspending non-essential equipment operation when an accidental fall is detected. Error correction engine: When the deviation between the actual EEG signal and the predicted value is greater than 15%, the system automatically adjusts the control parameters. The formula used is: Where L is the intention recognition loss function, calculated using cross-entropy, and η is the learning rate, with an initial value of 0.01 that decays with the number of iterations.

8. A multi-device smart home intent priority decision-making method based on electroencephalogram (EEG) signals according to claim 7, characterized in that, The error correction process of the error correction engine is as follows: S301: Deviation detection: Compare the actual status of the equipment with the expected value of the instruction every 50ms; S302: Root cause analysis: When the signal transmission delay > threshold, local caching is activated to accelerate the process; If the device is unresponsive, switch to the backup control channel; if the user's intent changes, recalculate the priority. S303: Parameter Adjustment: Perform short-term adjustments and long-term optimizations. Short-term adjustments modify the current control command parameters; long-term optimizations update the prediction model weights.

Images