Intelligent environment regulation method and device, storage medium and electronic device

Abstract

This application discloses an intelligent environment control method, device, storage medium, and electronic device. The intelligent environment control method includes collecting multimodal physiological signals from a user, including at least one of electroencephalogram (EEG), heart rate, and electrodermal conductance (EDC). It also involves adaptive preprocessing of the multimodal physiological signals to obtain target physiological data; extracting multidimensional features from the target physiological data and inputting these features into a dynamic benchmark model to obtain fatigue state indicators; generating multi-device collaborative control commands based on the current environmental state, user historical data, and fatigue state indicators through an intelligent decision-making model; and executing the multi-device collaborative control commands to adjust the environmental devices in a coordinated manner. This application can significantly improve the intelligence level and user experience of intelligent environment control systems.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of Internet of Things technology, and particularly to an intelligent environment regulation method, device, storage medium and electronic device, and especially to an environment regulation method combined with an intelligent window and door system. Background Art

[0002] With the rapid development of Internet of Things and artificial intelligence technologies, intelligent environment regulation systems are gradually penetrating into multiple fields such as home, office, and medical care, aiming to improve the comfort, health level and work efficiency of users in specific spaces through automated means. Such intelligent environment regulation systems usually achieve the linkage control of devices such as lights, curtains, and air conditioners based on environmental sensors (such as temperature and humidity, light, and air quality sensors) or user preset scenarios. Among them, intelligent windows and doors, as an important interface for the interaction between buildings and the environment, play a key role in ventilation, lighting, heat insulation and safety protection, and are an indispensable execution unit in the environment regulation system.

[0003] However, most of the current intelligent environment regulation systems still stay at the response to environmental physical parameters or the execution of fixed patterns, lacking the real-time perception and understanding of the user's own physiological state, with insufficient intelligence and difficult to achieve optimization with the core goal of the user's physical health and comfort. Especially in the aspect of window and door control, existing solutions often only perform window opening and closing operations based on preset time or simple indoor and outdoor temperature differences, without dynamically adjusting in combination with the user's physiological state (such as fatigue, attention level), which limits the potential of the window and door system in improving the indoor environmental quality. Summary of the Invention

[0004] The embodiments of the present application provide an intelligent environment regulation method, device, storage medium and electronic device, which can significantly improve the intelligence level and user experience of the intelligent environment regulation system, and are particularly suitable for multi-device collaborative control scenarios including intelligent windows and doors.

[0005] In a first aspect, the embodiments of the present application provide an intelligent environment regulation method, including: Collecting the user's multimodal physiological signals, where the multimodal physiological signals include at least one of electroencephalogram signals, heart rate signals and galvanic skin response signals; Performing adaptive preprocessing on the multimodal physiological signals to obtain target physiological data; Extracting multi-dimensional features from the target physiological data and inputting the multi-dimensional features into a dynamic benchmark model to obtain a fatigue state index; Generating a multi-device collaborative control instruction through an intelligent decision model based on the current environmental state, user historical data and the fatigue state index; Executing the multi-device collaborative control instruction to perform linkage adjustment on environmental devices.

[0006] In the intelligent environment control method provided in this application embodiment, the intelligent decision-making model includes a first decision sub-model and a second decision sub-model; the step of generating multi-device collaborative control instructions based on the current environmental state, user historical data, and the fatigue state index through the intelligent decision-making model includes: An input vector is constructed based on the current environmental state, user historical data, and the fatigue state index. The input vectors are input into the first decision sub-model and the second decision sub-model respectively to obtain preliminary adjustment suggestions for each environmental device and the expected utility value of each candidate action; Based on the expected utility value, derive target action suggestions; The preliminary adjustment suggestions and the target action suggestions are weighted, fused, and conflict-resolved to generate multi-device collaborative control commands.

[0007] In the intelligent environment control method provided in this application embodiment, the step of weighted fusion and conflict resolution of the preliminary adjustment suggestion and the target action suggestion to generate multi-device collaborative control instructions includes: Assign dynamic fusion weights to the preliminary adjustment recommendations and the expected utility values; Based on the dynamic fusion weights, the preliminary adjustment suggestions and the target action suggestions are fused and calculated to generate an initial fusion control strategy; Determine whether there is a conflict between the preliminary adjustment suggestion and the target action suggestion, and generate multi-device collaborative control instructions based on the determination result and the initial fusion control strategy.

[0008] In the intelligent environment control method provided in this application embodiment, the step of generating multi-device collaborative control instructions based on the judgment result and the initial fusion control strategy includes: If a conflict exists, the initial fusion control strategy is modified according to the preset conflict resolution rules to generate a control strategy to be verified. If there is no conflict, the initial fusion control strategy will be used as the control strategy to be verified. The control strategy to be verified is subjected to executability verification and smoothing processing to generate multi-device collaborative control instructions.

[0009] In the intelligent environment control method provided in this application embodiment, determining whether there is a conflict between the preliminary adjustment suggestion and the target action suggestion includes: Compare the preliminary adjustment recommendations with the target action recommendations for the control parameters of the same environmental equipment; If the difference between the control parameters is less than the adjustment resolution threshold of the environmental device and less than the user perception difference threshold, then it is determined that there is no conflict. Otherwise, it is determined that a conflict exists.

[0010] In the intelligent environment control method provided in this application embodiment, the step of assigning dynamic fusion weights to the preliminary adjustment suggestion and the target action suggestion includes: Obtain the first confidence level output by the first decision sub-model and the second confidence level output by the second decision sub-model; Calculate the matching degree between the input vector and the training data distribution in the first decision sub-model and the second decision sub-model; Based on the first confidence level, the second confidence level, and the matching degree, the dynamic fusion weights of the preliminary adjustment suggestions and the target action suggestions are determined by a preset weight allocation scheme.

[0011] In the intelligent environment regulation method provided in this application embodiment, the adaptive preprocessing of the multimodal physiological signals to obtain target physiological data includes: The multimodal physiological signals are timestamped to generate a synchronization signal sequence; The synchronization signal sequence is subjected to adaptive filtering and artifact suppression processing to generate a clean physiological signal; The clean physiological signals are subjected to quality assessment and segment selection to obtain target physiological data that meet preset quality standards.

[0012] Secondly, embodiments of this application provide an intelligent environmental control device, comprising: The signal acquisition unit is used to acquire the user's multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and electrodermal conductance signals. The signal processing unit is used to perform adaptive preprocessing on the multimodal physiological signals to obtain target physiological data; The feature extraction unit is used to extract multi-dimensional features from the target physiological data and input the multi-dimensional features into the dynamic benchmark model to obtain fatigue state indicators. The instruction generation unit is used to generate multi-device collaborative control instructions based on the current environmental state, user historical data, and the fatigue state index, through an intelligent decision-making model. The instruction execution unit is used to execute the multi-device collaborative control instructions and to perform coordinated adjustment of environmental equipment.

[0013] Thirdly, this application provides a storage medium storing a plurality of instructions that are applicable to a processor for loading to execute any of the intelligent environment control methods described above.

[0014] Fourthly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent environment control method described in any of the above claims.

[0015] In summary, the intelligent environment control method provided in this application includes: collecting multimodal physiological signals from a user, wherein the multimodal physiological signals include at least one of electroencephalogram (EEG) signals, heart rate signals, and electrodermal conductance (EDC) signals; performing adaptive preprocessing on the multimodal physiological signals to obtain target physiological data; extracting multidimensional features from the target physiological data and inputting the multidimensional features into a dynamic benchmark model to obtain fatigue state indicators; generating multi-device collaborative control instructions based on the current environmental state, user historical data, and the fatigue state indicators through an intelligent decision-making model; and executing the multi-device collaborative control instructions to perform coordinated adjustment of environmental devices. This application's embodiments introduce real-time perception and analysis of multimodal physiological signals and combine them with a dynamic benchmark model to achieve personalized fatigue state assessment. This fundamentally solves the problem that existing technologies cannot perceive the user's internal physiological state. Furthermore, by utilizing an intelligent decision-making model that integrates fuzzy logic and reinforcement learning, it generates multi-device collaborative control commands by combining the current environmental state with the user's historical data. This achieves a leap from "passively responding to environmental parameters or fixed modes" to "actively perceiving, understanding, and coordinating and adjusting," thereby truly optimizing the environment dynamically with the user's real-time health and comfort state as the core, significantly improving the intelligence level and user experience of the intelligent environmental control system. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram illustrating an application scenario of the intelligent environmental control method provided in the embodiments of this application.

[0018] Figure 2 This is a flowchart illustrating the intelligent environmental control method provided in the embodiments of this application.

[0019] Figure 3 This is a schematic diagram of the intelligent environmental control device provided in the embodiments of this application.

[0020] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0021] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0022] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of this application may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0023] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0024] In the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0025] In the description of this application, it should be noted that the terms "upper," "lower," "left," "right," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. In addition, terms such as "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0026] Most current intelligent environmental control systems remain at the level of responding to environmental physical parameters or executing fixed patterns, lacking real-time perception and understanding of the human body's physiological state. Their level of intelligence is insufficient, making it difficult to achieve true optimization centered on user health and comfort. Furthermore, existing door and window control systems are mostly based on simple time or temperature difference rules, failing to effectively link with user physiological states and other indoor equipment, thus limiting their potential to improve indoor environmental quality.

[0027] Based on this, embodiments of this application provide an intelligent environment control method, apparatus, storage medium, and electronic device. Specifically, the intelligent environment control apparatus can be integrated into an electronic device, which can be a server or a terminal, etc. The terminal can include mobile phones, wearable smart devices, tablets, laptops, and personal computers (PCs) and other computers and auxiliary devices. The server can be a single server or a server cluster composed of multiple servers, and can be a physical server or a virtual server.

[0028] For example, such as Figure 1 As shown, the electronic device can collect the user's multimodal physiological signals, including at least one of electroencephalogram (EEG), heart rate, and skin conductance signals; adaptively preprocess the multimodal physiological signals to obtain target physiological data; extract multidimensional features from the target physiological data and input the multidimensional features into a dynamic benchmark model to obtain fatigue state indicators; based on the current environmental state, user historical data, and fatigue state indicators, generate multi-device collaborative control commands through an intelligent decision-making model; execute the multi-device collaborative control commands to adjust the environmental devices in a coordinated manner. The environmental devices include smart doors and windows, smart lighting, and smart air conditioning, among which smart doors and windows can automatically adjust their opening angle, ventilation mode, and shading status according to commands. The technical solutions shown in this application will be described in detail below through specific embodiments. It should be noted that the order of description of the following embodiments is not intended to limit the priority of the embodiments.

[0029] Please see Figure 2 , Figure 2 This is a flowchart illustrating the intelligent environment control method provided in this application embodiment. The specific flow of the intelligent environment control method can be as follows: 101. Collect the user's multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and skin conductance signals.

[0030] In this embodiment, the user's real-time physiological signals can be collected synchronously through a variety of wearable or non-contact physiological sensors.

[0031] For example, in a home study setting, a user wears a lightweight headband with integrated electroencephalogram (EEG) electrodes, and a smart bracelet on their wrist that tracks heart rate (HR) and galvanic skin response (GSR) signals. When the user is reading or working for extended periods, these signals are continuously collected. EEG signals reflect the brain's electrical activity, heart rate signals characterize the excitation level of the autonomic nervous system, and GSR signals are related to emotional arousal and attention. Combining these three signals creates a multimodal physiological signal profile for assessing the user's physiological state, especially fatigue.

[0032] 102. Adaptive preprocessing is performed on multimodal physiological signals to obtain target physiological data.

[0033] In practical applications, the raw multimodal physiological signals inevitably contain various noises and interferences, which can lead to severe distortion in subsequent state analysis if used directly. Therefore, this embodiment uses adaptive preprocessing to transform the raw multimodal physiological signals into high-quality target physiological data.

[0034] In some embodiments, step 102 may specifically include the following steps: 1021. Timestamp alignment of multimodal physiological signals to generate synchronization signal sequences.

[0035] It is understandable that, since EEG signals, heart rate signals, and skin conductance signals come from different hardware devices, there are slight differences in their internal clocks, sampling rates, and data transmission delays.

[0036] Therefore, a high-precision timestamp (e.g., accurate to the millisecond level) based on a unified clock source can be added to each frame of data at the data acquisition or receiving end. By re-interpolating and aligning physiological signals from different hardware devices using these timestamps, strict synchronization of different physiological signals along the timeline can be ensured, forming a synchronized signal sequence. For example, in driver fatigue monitoring scenarios, after aligning the electrodermal signals from a head-mounted EEG device and a skin conductance sensor on the steering wheel using timestamps, the correlation between brain fatigue fluctuations and skin conductance responses within a specific heartbeat cycle can be accurately analyzed, providing a reliable temporal basis for comprehensive judgment.

[0037] 1022. Adaptive filtering and artifact suppression are applied to the synchronization signal sequence to generate a clean physiological signal.

[0038] In practice, targeted purification algorithms can be applied to address the interference characteristics of each physiological signal. Specifically: For EEG signals: First, a notch filter can be used to filter out 50Hz / 60Hz power frequency interference. Then, an adaptive algorithm (such as independent component analysis) is used to automatically identify and separate electrooculography (EOG) artifacts caused by blinking and eye movements, as well as electromyography (EMG) artifacts caused by neck and facial muscle tension, and remove them from the EEG signal. For example, when a user blinks frequently in an office setting due to thinking, the large-amplitude pulses caused by these blinks can be filtered out in real time, retaining clean physiological signals that reflect the true state of the brain.

[0039] For heart rate / electrodermal signals: Adaptive noise cancellation can be performed using a motion sensor-based reference signal. For example, when a smart bracelet on a user's wrist collects heart rate / electrodermal signals, if it detects that the user is typing or engaging in slight activity, its built-in accelerometer data will be used to estimate motion artifacts and subtract them from the heart rate / electrodermal signals, resulting in a more stable clean physiological signal. In some embodiments, adaptive noise cancellation can also be performed based on a combined reference of motion sensor and environmental event information. For example, when an indoor smart door and window sensor detects a sudden change in the state of doors and windows (such as suddenly opening a window), this event is flagged as a potential source of interference. This event signal can be used, along with accelerometer data, to estimate transient artifacts that may be caused in the electrodermal and heart rate signals by sudden changes in airflow, temperature sensations, or the user's subconscious bodily reactions (such as shivering from a draft), and subtracted from the physiological signal, resulting in a more stable clean physiological signal.

[0040] 1023. Perform quality assessment and segment selection on clean physiological signals to obtain target physiological data that meet preset quality standards.

[0041] In addition, since physiological signals are not reliable enough at all times, it is necessary to conduct real-time quality assessment of clean physiological signals to obtain a quality score.

[0042] Evaluation metrics may include signal strength and signal-to-noise ratio (to determine whether the signal is too weak or still dominated by noise), physiological rationality (e.g., whether the heart rate is within a reasonable resting range, such as 40-120 beats / minute), and interchannel consistency (for multichannel EEG signals, to determine whether the EEG signals of adjacent channels show reasonable spatial correlation in order to identify problems such as poor electrode contact).

[0043] In practical implementation, a quality scoring threshold can be set. Only signal segments in which the quality scores of all clean physiological signals are higher than this threshold within a continuous period (e.g., 2 seconds) will be selected as the final target physiological data. For example, in a home movie-watching scenario, when a user adjusts their posture, causing the EEG headset to temporarily shift, the system can identify a decrease in signal quality during that period and automatically exclude it from the analysis, avoiding the introduction of erroneous data that could interfere with the judgment of the user's relaxation or fatigue state.

[0044] This embodiment ensures that the data input to the subsequent feature extraction unit has a high degree of reliability.

[0045] 103. Extract multi-dimensional features from the target physiological data and input the multi-dimensional features into the dynamic benchmark model to obtain fatigue state indicators.

[0046] In some embodiments, a sliding time window (e.g., every 2 seconds as an analysis window with a step size of 1 second) can be used to analyze the target physiological data and calculate a series of features reflecting the physiological state to form a multi-dimensional feature vector.

[0047] This feature can include electroencephalogram (EEG) features, heart rate features, and skin conductance features. The specific extraction methods for these features are as follows: Electroencephalogram (EEG) characteristics: EEG characteristics are central to assessing central nervous system fatigue. EEG characteristics are obtained by calculating the power spectral density of the EEG signal in each frequency band within a sliding time window. The most typical EEG characteristic is the average power ratio (θ / α) of theta waves (4-8 Hz) to alpha waves (8-13 Hz). Studies have shown that as fatigue increases, theta wave activity, reflecting drowsiness, relatively increases, while alpha wave activity, reflecting relaxation and rest, may weaken or change, leading to an increase in this ratio. For example, in a late-night office setting, when users experience mental fatigue due to continuous work, the θ / α value in their EEG signals shows a significant upward trend.

[0048] Heart rate and skin conductance characteristics: While extracting EEG characteristics, time-domain indices of heart rate variability (HRV), such as the standard deviation of adjacent heartbeats, can be calculated from the heart rate signal. A decrease in this index is usually associated with accumulated stress and fatigue. From the skin conductance signal, the mean skin conductance level (SCL) or energy within a specific frequency band can be calculated. Decreased skin conductance activity may be associated with decreased attention and mental exhaustion.

[0049] Finally, features from different physiological dimensions are combined into a multi-dimensional feature vector, such as [θ / α: 2.1, HRV: 45 ms, mean SCL: 3.2μS]. This multi-dimensional feature vector can comprehensively characterize the user's physiological state within the sliding time window from multiple perspectives, such as brain cognition and autonomic nervous system arousal.

[0050] Because the extracted multi-dimensional features vary greatly among different users, directly using them for judgment lacks universality. Therefore, this embodiment introduces a "dynamic baseline model," the core function of which is to establish and maintain a personalized physiological baseline for each user. By comparing real-time features with the personal baseline, a personalized fatigue state index is calculated.

[0051] The dynamic benchmark model can be established as follows: ① Standardized Data Acquisition: In the implementation process, electronic devices can guide users into a preset "standard awake and relaxed state." For example, a smart speaker or mobile application can issue a voice prompt: "Please find a comfortable position to sit down, stay relaxed, and take a few minutes of natural breathing. Please try to avoid large movements or thinking about complex problems." At the same time, the electronic device can work with environmental devices to automatically adjust the lighting to a soft warm color temperature and play calm background white noise to create a stable calibration environment. Under this "standard awake and relaxed state," high-quality multimodal physiological signals are continuously collected for 3-5 minutes.

[0052] ② Construction of the personal characteristic benchmark database: Specifically, adaptive preprocessing and feature extraction can be performed on multimodal physiological signals first. Subsequently, statistical analysis is conducted on the extracted features to create a personalized profile for each feature.

[0053] First, for each feature (such as θ / α, HRV), its mean (μ) and standard deviation (σ) can be calculated over the entire calibration period. The mean represents the "normal central value" for that feature for the user, while the standard deviation describes its normal range of fluctuation. For example, the baseline for "θ / α" calculated for user A is: mean μ_θ / α = 1.5, standard deviation σ_θ / α = 0.2.

[0054] Next, the correlations between different features are analyzed, such as calculating the covariance matrix of θ / α and HRV. This allows the dynamic benchmark model to consider not only the deviation of individual features but also whether the feature combination pattern is abnormal when making subsequent judgments, thereby constructing a multi-dimensional personal benchmark cloud map. In some embodiments, a "normal state hyperellipsoid" centered on the mean of each feature and bounded by several times the standard deviation can be defined to form the initial dynamic benchmark model.

[0055] ③ Initialization of confidence level and update mechanism: After forming the initial dynamic baseline model, an initial confidence score can be assigned to it. Simultaneously, initial update parameters can be set. For example: Update weight (α): Determines the degree of impact of new data on the old benchmark (e.g., α=0.05 indicates that the new data has a small impact on the benchmark update, keeping the benchmark stable).

[0056] Learning trigger threshold: Defines the conditions under which automatic incremental updates can be triggered (such as detecting similar new "wake-up" patterns for several consecutive days).

[0057] This embodiment successfully constructs a quantitative dynamic benchmark model for users. This dynamic benchmark model not only records the physiological fingerprint of the user's "awake state," but also provides a precise mathematical scale and personalized basis for subsequent real-time calculation of "how far the current state deviates from the normal state" (i.e., fatigue state index).

[0058] In practical applications, multi-dimensional feature vectors can be compared with the user's latest personalized baseline stored in the dynamic benchmark model, thereby generating fatigue state indicators based on the comparison results.

[0059] In some embodiments, an algorithm that integrates multiple distance metrics can be used to calculate the deviation between the multi-dimensional feature vector and the personalized baseline. For example, the Mahalanobis distance between the multi-dimensional feature vector and the personalized baseline can be calculated. This Mahalanobis distance takes into account the correlation between features and their respective variances, and can more accurately measure the overall distance between the user's current state and the personalized baseline.

[0060] Next, the Mahalanobis distance is mapped to an intuitive fatigue state index using a preset S-curve function. This fatigue state index is typically designed as a scalar value between 0 and 100, with higher values ​​indicating deeper fatigue.

[0061] Finally, the fatigue status index and confidence level can be output, for example, {fatigue status index: 72, confidence level: 0.88}. This indicates that the current user's overall physiological state deviates from its personalized baseline by a score of 72 (belonging to "moderate to severe fatigue"), and the confidence level of this judgment is 88%.

[0062] For example, in a long-distance driving scenario, the driver initially has a good condition with a fatigue state index of 25. After driving continuously for two hours, the calculated θ / α ratio increased by 80% compared to the personalized baseline, HRV decreased by 40%, and skin conductance activity significantly weakened. The dynamic baseline model can integrate these deviations, outputting a fatigue state index of 78.

[0063] 104. Based on the current environmental conditions, user historical data, and fatigue status indicators, generate multi-device collaborative control commands through an intelligent decision-making model.

[0064] In this embodiment, the intelligent decision-making model includes a first decision sub-model and a second decision sub-model. Step 104 may include the following steps: 1041. Construct an input vector based on the current environmental state, user historical data, and fatigue state indicators.

[0065] In the specific implementation process, the current environmental status can be obtained through environmental sensors. For example, the light sensor reading is 300 lux (slightly bright), the temperature and humidity sensor shows a temperature of 24℃ and a humidity of 50%, and the time module provides information such as "Tuesday, 21:30". The door and window sensors provide status information: the current window opening angle is 45°, and the ventilation mode is natural ventilation.

[0066] User history data retrieved from user profiles includes, for example, showing that in the past month, whenever the user's fatigue index exceeded 60, 80% of their interaction records indicated that they ultimately adjusted the lights to a brightness level below 30%; and their preferred air conditioning temperature for summer nights was set to 26°C. User history data also shows that when fatigue levels are high, users tend to adjust the window opening angle to below 15° to reduce external disturbances.

[0067] 1042. Input the input vectors into the first decision sub-model and the second decision sub-model respectively to obtain preliminary adjustment suggestions for each environmental device and the expected utility value of each candidate action.

[0068] The first decision sub-model is a rule-based reasoning model based on fuzzy logic. This sub-model incorporates an "IF-THEN" fuzzy rule base defined by expert knowledge. It can convert precise values ​​in the input vector (e.g., "fatigue state index 72") into fuzzy linguistic values ​​(e.g., "fatigue state index high") and then match them with rules. Successfully matched rules trigger corresponding preliminary adjustment suggestions, which are a set of explicit but potentially ambiguous equipment instructions.

[0069] For example, a rule might be: "IF time is 'night' AND fatigue status index is 'high' AND ambient light is 'bright', THEN the light action is 'significantly dimmed', the curtain action is 'closed', the air conditioning mode is 'switch to sleep mode', and the window action is 'reduce opening angle to slight ventilation'..." The first decision sub-model would output preliminary adjustment suggestions such as {lights: dim to 25%, curtains: close, air conditioning: set to sleep mode, window: adjust opening angle to 10%}.

[0070] The second decision sub-model is an adaptive policy model based on reinforcement learning. This sub-model learns the long-term value of various adjustment actions under different states through long-term interaction with the user and environment. Internally, it maintains a "state-action value function." For the current state represented by the input vector, the second decision sub-model can evaluate all possible candidate actions (such as adjusting the lights to 10%, 20%, 30%, etc.; opening / closing the windows at 0%, 10%, 20%, etc.) and calculate the expected utility value of each candidate action. This expected utility value predicts the cumulative gain in user comfort over a future period after executing the candidate action.

[0071] 1043. Derive target action suggestions based on the expected utility value.

[0072] This embodiment transforms the expected utility value of the second decision sub-model into actionable target suggestions. The specific derivation process is as follows: ① Data structure for parsing expected utility values: The output of the second decision sub-model is a list or vector containing multiple candidate actions and their corresponding expected utility values. Each candidate action represents a specific control setting for one or more environmental devices. The expected utility value is a numerical value that quantifies the cumulative positive return that can be obtained in a future time window (such as the next 30 minutes) after executing the candidate action. This cumulative positive return typically integrates multiple objectives such as a decrease in fatigue status indicators, energy consumption costs, and user preference satisfaction.

[0073] For example, candidate action A: {light: {brightness: 20%, color temperature: 2700K}, air conditioner: {mode: air supply, temperature: 26℃}, window: {opening angle: 15%, mode: micro-ventilation}}, expected utility value: 8.5; Candidate Action B: {Lighting: {Brightness: 30%, Color Temperature: 3000K}, Air Conditioning: {Mode: Cooling, Temperature: 25℃}, Window: {Opening Angle: 5%, Mode: Closed}}, Expected Utility: 7.2; Candidate action C: {Lighting: {Brightness: 15%, Color Temperature: 2200K}, Air Conditioning: {Mode: Sleep, Temperature: 26℃}, Window: {Opening Angle: 0%, Mode: Fully Closed}}, Expected Utility: 9.1.

[0074] ② Determine target action suggestions: In some embodiments, a preset strategy can be used to determine the target action suggestion from the candidate action list. For example, a single-objective optimization strategy can be used, which selects the candidate action with the highest expected utility value as the target action suggestion. In the example above, candidate action C has the highest expected utility value (9.1), and is therefore determined as the target action suggestion.

[0075] For example, in an office setting, when a user is determined to be in a state of "afternoon distraction," the second sub-decision model might assess that "briefly increasing the light color temperature to 5000K (cool white light)" has the highest expected utility value, because historical data shows that this candidate action can effectively increase the user's alertness. Therefore, this candidate action is deduced as the target action suggestion. Similarly, in a residential setting, when a user is detected to have entered a deep sleep state, the second sub-decision model might assess that "completely closing the windows and starting the fresh air system" has the highest expected utility value, in order to maintain stable indoor temperature and humidity and reduce external noise interference.

[0076] Based on the candidate action C above, the target action suggestion can be formatted as: {Target action suggestion:{Device:“Smart Light”,Parameters:{Brightness:15%,Color Temperature:2200K},{Device:“Smart Air Conditioner”,Parameters:{Mode:“Sleep”,Temperature:26℃}},{Device:“Smart Window”,Parameters:{Opening Angle:0%,Mode:“Fully Closed”}}}.

[0077] Understandably, the target action suggestion itself is a solution that takes into account the interaction between environmental devices. For example, it may suggest "dimming the lights" and "turning on the air conditioner's sleep mode" simultaneously, rather than adjusting a single environmental device in isolation, reflecting a collaborative control concept based on overall benefits.

[0078] 1044. The preliminary adjustment suggestions and target action suggestions are weighted, integrated, and conflict-resolved to generate multi-device collaborative control instructions.

[0079] It is understandable that the outputs of the two decision sub-models may be consistent or differ. Therefore, a weighted fusion and conflict resolution mechanism can be used to reach a final decision. In some embodiments, step 1043 may include the following steps: 1044a. Assign dynamic fusion weights to the initial adjustment recommendations and target action recommendations.

[0080] Specifically, the system can obtain the first confidence level of the first decision sub-model output and the second confidence level of the second decision sub-model output; calculate the matching degree between the input vector and the training data distribution in the first and second decision sub-models; and determine the dynamic fusion weights of the preliminary adjustment suggestions and the target action suggestions based on the first confidence level, the second confidence level, and the matching degree through a preset weight allocation scheme.

[0081] For the first decision sub-model, its first confidence score is typically calculated based on the average confidence level or the combined trigger strength of the currently activated fuzzy rules. For example, if the input vector (such as "late night, high fatigue, strong light") can simultaneously and strongly activate multiple fuzzy rules with consistent conclusions, then the first decision sub-model will output a high first confidence score (e.g., 0.9). Conversely, if the rule match is fuzzy or triggers a small number of contradictory rules, the first confidence score will be low (e.g., 0.4).

[0082] For the second decision sub-model, its second confidence level reflects the certainty of its value assessment. If, for the current state, the expected utility values ​​of all candidate actions differ significantly (large variance), it indicates that the second decision sub-model is very certain which candidate action is optimal, and thus the second confidence level is high. If the utility values ​​of all candidate actions are very similar (small variance), it indicates that the second decision sub-model is less certain in the current state, and thus the second confidence level is low.

[0083] Understandably, confidence level is key to weight allocation. The core logic is: the more confident a decision sub-model is in its judgment under the current state, the greater its weight should be in the final decision. Even if a decision sub-model has a high degree of fit with the current scenario, if its output has low confidence, its weight will decrease, because low confidence suggests its output may be unreliable. Conversely, even if a decision sub-model has only a moderate degree of scenario fit, if it gives a high-confidence output based on sufficient internal evidence, its weight will be increased accordingly.

[0084] In some embodiments, the matching degree can be determined by evaluating the similarity between the input vector and the training data distribution of each decision sub-model. For example, Mahalanobis distance or kernel-based methods can be used to determine whether the current state is closer to a typical scenario covered by the first decision sub-model or closer to a region that the second decision sub-model has fully explored through historical interactions.

[0085] Based on the first confidence level, the second confidence level, and the matching degree, the dynamic fusion weights of the preliminary adjustment suggestions and the target action suggestions can be determined by a preset weight allocation scheme.

[0086] For example, during the afternoon of a weekday, a unique "afternoon fatigue" pattern was detected in users. This pattern had been thoroughly learned by the second decision sub-model from historical data, resulting in a high degree of matching between the second decision sub-model and the input vector. However, for the first decision sub-model, this might be a less well-defined and atypical scenario, thus the matching degree between the first decision sub-model and the input vector would be moderate.

[0087] At this point, the second decision sub-model not only has a high matching degree, but also outputs significantly different expected utility values ​​because it identifies clear individual patterns. Therefore, its second confidence level is also very high (e.g., 0.85). In contrast, the first decision sub-model, due to its atypical scenario, may not have complete rule triggering, so its first confidence level is only moderate (e.g., 0.6).

[0088] The preset weight allocation scheme integrates the second decision sub-model with "high matching degree + high confidence" and the first decision sub-model with "medium matching degree + medium confidence" to assign corresponding weights to the preliminary adjustment suggestions and the target action suggestions. For example, the second decision sub-model may receive a higher dynamic fusion weight (e.g., 0.7) for its target action suggestions due to its dual advantages in matching degree and confidence, while the preliminary adjustment suggestions output by the first decision sub-model may receive a relatively lower weight (e.g., 0.3).

[0089] 1044b. Based on dynamic fusion weights, the preliminary adjustment suggestions and target action suggestions are fused and calculated to generate an initial fusion control strategy.

[0090] Specifically, preliminary adjustment suggestions and target action suggestions can be merged based on dynamic fusion weights. For example, a weighted average can be applied to the adjustment parameters of each environmental device.

[0091] For example, assuming the initial adjustment suggestion of the first decision sub-model for the target air conditioning temperature is 26℃ (weight 0.6), and the target action suggestion of the second sub-model is 25℃ (weight 0.4), then the air conditioning temperature in the initial fusion control strategy might be calculated as 26 * 0.6 + 25 * 0.4 = 25.6℃. For the window opening angle, the initial adjustment suggestion of the first decision sub-model is 10% (weight 0.6), and the target action suggestion of the second sub-model is 0% (weight 0.4), then the window opening angle in the initial fusion control strategy might be calculated as 10% * 0.6 + 0% * 0.4 = 6%. By repeating this process for all environmental devices, a complete initial fusion control strategy can be generated.

[0092] 1044c. Determine whether there is a conflict between the preliminary adjustment suggestion and the target action suggestion, and generate multi-device collaborative control instructions based on the judgment result and the initial fusion control strategy.

[0093] In practical applications, strict conflict detection and resolution are required to ensure consistency within the strategy.

[0094] Specifically, the initial adjustment suggestions and the target action suggestions are compared for the control parameters of the same environmental device. If the difference between the control parameters is less than the adjustment resolution threshold of the environmental device (such as a 1% brightness step for a smart bulb or a 1% opening angle step for a smart window) and less than the user perception difference threshold (such as a 5% brightness change that the human eye cannot reliably distinguish or a 5% opening angle change corresponding to the human body's perception difference threshold for a light breeze), then it is determined that there is no conflict; otherwise, it is determined that there is a conflict.

[0095] For example, regarding the opening / closing degree of curtains, the initial adjustment suggestion is "closed (0%)", and the target action suggestion is "leave a 10% gap". If the environmental device resolution is 1%, but the user's perception threshold is measured to be 5%, then the difference of 10% is much greater than both, so it is determined to be a conflict. Similarly, regarding the opening / closing angle of windows, the initial adjustment suggestion is "15%", and the target action suggestion is "5%", the difference is 10%, which is greater than both the device resolution of 1% and the user's perception threshold of 5%, so it is determined to be a conflict.

[0096] In the specific implementation process, if a conflict exists, the initial fusion control strategy will be modified according to the preset conflict resolution rules to generate a control strategy to be verified; if there is no conflict, the initial fusion control strategy will be used as the control strategy to be verified.

[0097] For example, the preset conflict resolution rules can stipulate that "conflicts involving safety, energy conservation, or equipment protection will be prioritized based on the initial adjustment suggestions output by the first decision sub-model," and "conflicts involving personalized comfort habits will be prioritized based on the target action suggestions output by the second decision sub-model." After modifying the initial fusion control strategy according to the preset conflict resolution rules, the control strategy to be verified is obtained. For conflicts involving doors and windows, if safety is involved (such as when a wind and rain sensor detects severe weather), the closing suggestion from the first decision sub-model will be prioritized; if only ventilation preferences are involved, the learning suggestion from the second decision sub-model will be prioritized.

[0098] Finally, the control strategy to be verified undergoes executability verification (checking whether it exceeds the capabilities of the environmental equipment) and smoothing processing (converting abrupt instructions into a gradual instruction sequence) to generate multi-device collaborative control instructions. For example, "immediately adjust the lights from 80% to 25%" is transformed into "gradually dim to 25% in three stages over 12 seconds." "Immediately close the windows from 45° to 0°" is transformed into "gradually close them in stages over 30 seconds to avoid sudden static pressure changes and noise." Ultimately, a set of multi-device collaborative control instructions is output and sent to each environmental device for execution.

[0099] 105. Execute multi-device collaborative control commands to adjust environmental equipment in a coordinated manner.

[0100] Specifically, the timing can be carefully orchestrated based on the physical characteristics of the devices and the user experience (e.g., first initiating a slow curtain closure, then triggering a synchronized dimming of the lights, and finally adjusting the air conditioning mode while simultaneously controlling the windows to gradually adjust to the target opening angle). Control commands are then issued and monitored through IoT protocols. Ultimately, this achieves smooth integration and overall linkage of the actions of environmental devices such as curtains, lights, air conditioning, and windows, thereby completing a natural and coherent transition from "bright office" to "dim rest" environments, creating a seamless, comfortable, and highly collaborative immersive experience for users.

[0101] The intelligent environmental control method provided in this application can be applied to various scenarios. For example: Home Theater Mode Activation: When user fatigue and a desire to relax are detected, the following actions may be triggered: 1) The main lights gradually dim and turn off within 15 seconds; 2) The blackout curtains close completely simultaneously; 3) The projection screen slowly lowers; 4) The surround sound system turns on and plays soft background music; 5) The smart windows automatically close to 0%, isolating external light and noise. The entire process takes approximately 20 seconds, creating an immersive viewing atmosphere.

[0102] Natural wake-up in the morning: Based on the user's sleep cycle and schedule, the system activates 30 minutes before the scheduled wake-up time: 1) Curtains slowly open to 60%, gradually introducing natural light; 2) Bedroom lights simulate sunrise, with color temperature gradually changing from 2700K to 4000K and brightness gradually increasing from 0% to 40%; 3) The air conditioner is pre-adjusted to a comfortable daytime temperature; 4) The smart speaker begins playing soft music or news briefings; 5) The smart window slowly opens to 20%, introducing fresh air and promoting natural wake-up. This coordinated regulation helps users achieve a painless and natural wake-up.

[0103] Intelligent ventilation scenario: When the indoor air quality sensor detects an increase in CO2 concentration and the user's fatigue status indicator shows mild fatigue (suitable for brief wake-up adjustment), the system can perform the following actions: 1) The intelligent window slowly opens to 30% to create airflow; 2) The air conditioner switches to energy-saving air supply mode; 3) If the outdoor noise is loud, the window opening angle is automatically optimized to 15° and silent mode is activated; 4) After adjustment, the light color temperature is finely adjusted to fresh mode to help the user restore energy.

[0104] In summary, the intelligent environment control method provided in this application includes: collecting multimodal physiological signals from the user, including at least one of electroencephalogram (EEG), heart rate, and electrodermal conductance (EDA); adaptively preprocessing the multimodal physiological signals to obtain target physiological data; extracting multidimensional features from the target physiological data and inputting the multidimensional features into a dynamic benchmark model to obtain fatigue state indicators; generating multi-device collaborative control commands through an intelligent decision-making model based on the current environmental state, user historical data, and fatigue state indicators; and executing the multi-device collaborative control commands to perform coordinated adjustment of environmental devices. This application, by introducing real-time perception and analysis of multimodal physiological signals and combining them with a dynamic benchmark model to achieve personalized fatigue state assessment, fundamentally solves the problem that existing technologies cannot perceive the user's internal physiological state. Furthermore, this application utilizes an intelligent decision-making model that integrates fuzzy logic and reinforcement learning. By combining the current environmental state with historical user data, it generates multi-device collaborative control commands, achieving a leap from "passively responding to environmental parameters or fixed modes" to "actively sensing, understanding personally, and collaboratively adjusting." This truly optimizes the environment dynamically based on the user's real-time health and comfort, significantly improving the intelligence level and user experience of the intelligent environmental control system. In particular, this application incorporates the intelligent door and window system as the core execution unit into the collaborative control framework, realizing intelligent linkage between doors and windows and lighting, air conditioning, and other equipment. This plays a crucial role in optimizing indoor air quality, regulating natural lighting, and reducing energy consumption, providing users with a healthier, more comfortable, and energy-efficient indoor environment experience.

[0105] To facilitate better implementation of the intelligent environment control method provided in this application, this application also provides an intelligent environment control device. The meanings of the terms used are the same as in the intelligent environment control method described above, and specific implementation details can be found in the descriptions within the method embodiments.

[0106] Please see Figure 3 , Figure 3 This is a schematic diagram of the intelligent environment control device provided in an embodiment of this application. The intelligent environment control device may include a signal acquisition unit 201, a signal processing unit 202, a feature extraction unit 203, an instruction generation unit 204, and an instruction execution unit 205. The signal acquisition unit 201 is used to acquire the user's multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and skin conductance signals. Signal processing unit 202 is used to perform adaptive preprocessing on multimodal physiological signals to obtain target physiological data; The feature extraction unit 203 is used to extract multi-dimensional features from the target physiological data and input the multi-dimensional features into the dynamic benchmark model to obtain fatigue state indicators. The instruction generation unit 204 is used to generate multi-device collaborative control instructions based on the current environmental status, user historical data and fatigue status indicators, through an intelligent decision-making model. The instruction execution unit 205 is used to execute multi-device collaborative control instructions and to coordinate and adjust environmental equipment.

[0107] For specific implementation methods of each of the above units, please refer to the embodiments of the intelligent environment control method described above, which will not be repeated here.

[0108] In summary, the intelligent environmental control device provided in this application embodiment can acquire multimodal physiological signals from the user through the signal acquisition unit 201. These multimodal physiological signals include at least one of electroencephalogram (EEG), heart rate, and skin conductance signals. The signal processing unit 202 performs adaptive preprocessing on the multimodal physiological signals to obtain target physiological data. The feature extraction unit 203 extracts multidimensional features from the target physiological data and inputs these features into a dynamic benchmark model to obtain fatigue state indicators. The instruction generation unit 204 generates multi-device collaborative control instructions based on the current environmental state, user historical data, and fatigue state indicators through an intelligent decision-making model. The instruction execution unit 205 executes the multi-device collaborative control instructions to perform coordinated adjustment of the environmental devices. This application embodiment, by introducing real-time perception and analysis of multimodal physiological signals and combining them with a dynamic benchmark model to achieve personalized fatigue state assessment, fundamentally solves the problem that existing technologies cannot perceive the user's internal physiological state. Furthermore, this application embodiment utilizes an intelligent decision-making model that integrates fuzzy logic and reinforcement learning, combining the current environmental state with user historical data to generate multi-device collaborative control commands. This achieves a leap from "passively responding to environmental parameters or fixed modes" to "actively perceiving, understanding personally, and coordinating and adjusting," thereby truly optimizing the environment dynamically with the user's real-time health and comfort as the core, significantly improving the intelligence level and user experience of the intelligent environmental control system.

[0109] This application also provides an electronic device that may integrate the intelligent environmental control device of this application embodiment, such as... Figure 4 As shown, it illustrates a structural schematic diagram of the electronic device involved in the embodiments of this application, specifically: The electronic device may include components such as a processor 301 with one or more processing cores and a memory 302 with one or more computer-readable storage media. Those skilled in the art will understand that... Figure 4 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 301 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs stored in the memory 302 and / or this application, and by calling data stored in the memory 302, it performs various functions and processes data, thereby providing overall monitoring of the electronic device. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operation of the storage medium, user interface, and application programs, while the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 301.

[0110] The memory 302 can be used to store software programs and this application. The processor 301 executes various functional applications and data processing by running the software programs and this application stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store applications required for operating the storage medium and at least one function; the data storage area may store data created based on the use of the electronic device. In addition, the memory 302 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0111] Although not shown, the electronic device may also include a display unit, an input unit, and a power supply, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 301 in the electronic device loads the executable files corresponding to the processes of one or more application programs into the memory 302 according to the following instructions, and the processor 301 runs the application programs stored in the memory 302 to realize various functions, as follows: Collect users' multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and skin conductance signals. Adaptive preprocessing is performed on multimodal physiological signals to obtain target physiological data; Multi-dimensional features are extracted from the target physiological data and input into the dynamic benchmark model to obtain fatigue state indicators; Based on the current environmental conditions, user historical data, and fatigue status indicators, a multi-device collaborative control command is generated through an intelligent decision-making model. It executes multi-device collaborative control commands to coordinate and adjust environmental equipment.

[0112] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0113] Therefore, embodiments of this application provide a storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the methods provided in embodiments of this application. For example, the instructions can execute the following steps: Collect users' multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and skin conductance signals. Adaptive preprocessing is performed on multimodal physiological signals to obtain target physiological data; Multi-dimensional features are extracted from the target physiological data and input into the dynamic benchmark model to obtain fatigue state indicators; Based on the current environmental conditions, user historical data, and fatigue status indicators, a multi-device collaborative control command is generated through an intelligent decision-making model. It executes multi-device collaborative control commands to coordinate and adjust environmental equipment.

[0114] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0115] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0116] Since the instructions stored in the storage medium can execute the steps of any method provided in the embodiments of this application, the beneficial effects that any method provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0117] The above provides a detailed description of the intelligent environmental control method, device, storage medium, and electronic device provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of this application. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligent environmental control, characterized in that, include: Collect the user's multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and electrodermal conductance signals; The multimodal physiological signals are adaptively preprocessed to obtain the target physiological data; Multidimensional features are extracted from the target physiological data and input into the dynamic benchmark model to obtain fatigue state indicators; Based on the current environmental conditions, user historical data, and the fatigue state indicators, a multi-device collaborative control command is generated through an intelligent decision-making model. The multi-device collaborative control command is executed to adjust the environmental equipment in a coordinated manner.

2. The intelligent environmental control method as described in claim 1, characterized in that, The intelligent decision-making model includes a first decision sub-model and a second decision sub-model; The process of generating multi-device collaborative control commands based on the current environmental state, user historical data, and the fatigue state index through an intelligent decision-making model includes: An input vector is constructed based on the current environmental state, user historical data, and the fatigue state index. The input vectors are input into the first decision sub-model and the second decision sub-model respectively to obtain preliminary adjustment suggestions for each environmental device and the expected utility value of each candidate action; Based on the expected utility value, derive target action suggestions; The preliminary adjustment suggestions and the target action suggestions are weighted, fused, and conflict-resolved to generate multi-device collaborative control commands.

3. The intelligent environmental control method as described in claim 2, characterized in that, The step of weighted fusion and conflict resolution of the preliminary adjustment suggestions and the target action suggestions to generate multi-device collaborative control commands includes: Assign dynamic fusion weights to the preliminary adjustment suggestions and the target action suggestions; Based on the dynamic fusion weights, the preliminary adjustment suggestions and the target action suggestions are fused and calculated to generate an initial fusion control strategy; Determine whether there is a conflict between the preliminary adjustment suggestion and the target action suggestion, and generate multi-device collaborative control instructions based on the determination result and the initial fusion control strategy.

4. The intelligent environmental control method as described in claim 3, characterized in that, The step of generating multi-device collaborative control instructions based on the judgment result and the initial fusion control strategy includes: If a conflict exists, the initial fusion control strategy is modified according to the preset conflict resolution rules to generate a control strategy to be verified. If there is no conflict, the initial fusion control strategy will be used as the control strategy to be verified. The control strategy to be verified is subjected to executability verification and smoothing processing to generate multi-device collaborative control instructions.

5. The intelligent environmental control method as described in claim 3, characterized in that, The determination of whether there is a conflict between the preliminary adjustment suggestion and the target action suggestion includes: Compare the preliminary adjustment recommendations with the target action recommendations for the control parameters of the same environmental equipment; If the difference between the control parameters is less than the adjustment resolution threshold of the environmental device and less than the user perception difference threshold, then it is determined that there is no conflict. Otherwise, it is determined that a conflict exists.

6. The intelligent environmental control method as described in claim 3, characterized in that, Assigning dynamic fusion weights to the preliminary adjustment suggestions and the target action suggestions includes: Obtain the first confidence level output by the first decision sub-model and the second confidence level output by the second decision sub-model; Calculate the matching degree between the input vector and the training data distribution in the first decision sub-model and the second decision sub-model; Based on the first confidence level, the second confidence level, and the matching degree, the dynamic fusion weights of the preliminary adjustment suggestions and the target action suggestions are determined by a preset weight allocation scheme.

7. The intelligent environmental control method as described in claim 1, characterized in that, The adaptive preprocessing of the multimodal physiological signals to obtain target physiological data includes: The multimodal physiological signals are timestamped to generate a synchronization signal sequence; The synchronization signal sequence is subjected to adaptive filtering and artifact suppression processing to generate a clean physiological signal; The clean physiological signals are subjected to quality assessment and segment selection to obtain target physiological data that meet preset quality standards.

8. An intelligent environmental control device, characterized in that, include: The signal acquisition unit is used to acquire the user's multimodal physiological signals, which include at least one of electroencephalogram (EEG) signals, heart rate signals, and electrodermal conductance signals. The signal processing unit is used to perform adaptive preprocessing on the multimodal physiological signals to obtain target physiological data; The feature extraction unit is used to extract multi-dimensional features from the target physiological data and input the multi-dimensional features into the dynamic benchmark model to obtain fatigue state indicators. The instruction generation unit is used to generate multi-device collaborative control instructions based on the current environmental state, user historical data, and the fatigue state index, through an intelligent decision-making model. The instruction execution unit is used to execute the multi-device collaborative control instructions and to perform coordinated adjustment of environmental equipment.

9. A storage medium, characterized in that, The storage medium stores multiple instructions, which are applicable to the processor for loading to execute the intelligent environment control method according to any one of claims 1-7.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent environment control method as described in any one of claims 1-7.

Images