EEG feedback optimization method and system for building appearance visual features and emotion regulation

Abstract

The invention relates to an EEG feedback optimization method and system for building appearance visual features and emotion regulation, and the method comprises the steps: constructing a building appearance 3D model matched with the building features of a target region, and carrying out the fine processing, and obtaining a 3D model library of a plurality of visual stimulation sources; constructing a virtual scene comprising a building appearance 3D model and at least one user, and configuring the view angle and layout of the corresponding user in the virtual scene according to the action, implementation and behavior mode of the participant in the real environment; obtaining EEG signals of participants and extracting emotional characteristic indexes; based on the predicted emotional state of the participant under the current building appearance, the influence of the building appearance on the emotional pressure and the positive emotion is analyzed; and on the basis of the current building appearance 3D model parameters, predicting an emotion regulation effect corresponding to the parameters so as to minimize the difference between the predicted emotion regulation effect and a preset target, and iterating the parameters of the building appearance 3D model.

Description

Technical Field

[0001] The present invention relates to the field of neuroscience and architectural design technology, and in particular to an EEG feedback optimization method and system for architectural appearance visual features and emotion regulation. Background Art

[0002] With the rapid advancement of urbanization, urban residents are facing unprecedented emotional pressure while enjoying the convenience of modern life among high-rise buildings. The urban environment, especially the appearance of buildings, as an indispensable visual language under the city skyline, is no longer limited to functional and aesthetic considerations. It also needs to incorporate care for the mental health of residents. A well-designed architectural appearance can create a warm, peaceful or inspiring atmosphere with its unique color matching, form composition and light and shadow effects, which subtly regulates the emotional state of residents and becomes a positive factor in alleviating the pressure of urban life and promoting mental health.

[0003] Currently, the exploration of the relationship between the visual features of building appearance and the emotional changes of residents mainly relies on methods such as subjective questionnaires and facial expression analysis. Although these methods have revealed the potential connection between the two to a certain extent, their limitations cannot be ignored. Subjective questionnaires are easily affected by the respondents' personal bias, memory bias and immediate emotional state, resulting in highly subjective results and difficulty in forming generally applicable conclusions. Although facial expression analysis can capture immediate emotional reactions, it is difficult to penetrate the surface and deeply explore the complex psychological mechanisms behind emotions and changes in the physiological state of the brain. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide an EEG feedback optimization method and system for building appearance visual features and emotion regulation, which evaluates the impact of building appearance on emotional state based on EEG signals and optimizes the visual features of the building appearance through feedback.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] One aspect of the present invention provides an EEG feedback optimization method for building appearance visual features and emotion regulation, comprising the following steps:

[0007] Step S1, constructing a 3D model of the building appearance that matches the building characteristics of the target area and performing fine processing to obtain a 3D model library of multiple visual stimulus sources;

[0008] Step S2, constructing a virtual scene including a 3D model of the building appearance and at least one user, and configuring the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations and behavior patterns of the participants in the real environment;

[0009] Step S3, obtaining the EEG signal of the participant and extracting the emotion characteristic index;

[0010] Step S4, analyzing the influence of the building appearance on emotional stress and positive emotions based on the predicted emotional state of the participants under the current building appearance;

[0011] Step S5, based on the current building appearance 3D model parameters, predict the emotion regulation effect corresponding to the parameters, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and iterate the parameters of the building appearance 3D model.

[0012] As a preferred technical solution, the step S1 includes:

[0013] Step S101, obtaining one or more of local documents, historical records, and resident interview data, and performing data cleaning, deduplication, and classification processing to extract cultural, color, and style information of the building appearance;

[0014] Step S102, analyzing the influence of different climate factors and terrain factors on the building appearance, obtaining the climate and terrain information of the building location, and integrating it with the culture, color and style information of the building appearance to obtain a preliminary designed 3D model of the building appearance;

[0015] Step S103, based on the preliminary designed 3D model of the building appearance, the color of the model is refined by considering the harmony and unity of the overall color tone and the effect of color changes under different materials and lighting conditions, and the material of the model is refined by 3D iterative rendering;

[0016] Step S104: Classify and store the refined building appearance 3D models in a 3D model library.

[0017] As a preferred technical solution, the step S2 includes:

[0018] Step S201, obtaining GIS data and real-world photos, constructing a virtual scene including a 3D model of the building appearance and at least one user, adjusting the perspective of the corresponding user in the virtual scene according to the actions, sight lines and behavior patterns of the participants in the real environment, dynamically adjusting the scene layout according to the behavior patterns of the participants in the real environment, and predicting and simulating weather conditions and lighting changes in the virtual scene;

[0019] Step S202, obtaining the operation data of each participant in real time, updating the user's behavior pattern and the interaction intention between users, and detecting and resolving the behavior conflicts between users.

[0020] As a preferred technical solution, the step S3 includes:

[0021] Step S301, obtaining the electrical signal of the EEG sensor, performing analog-to-digital conversion and signal enhancement based on automatic gain control, and obtaining the EGG signal of the participant through noise suppression and artifact removal;

[0022] Step S302 , performing multi-scale decomposition on the EGG signal to obtain a multi-band characteristic waveform, and retaining a feature subset related to the emotional state as an emotional feature indicator through feature screening.

[0023] As a preferred technical solution, the step S4 includes:

[0024] Step S401, mapping the emotion characteristic index to a pre-constructed emotion score quantification system to obtain a score reflecting the emotional state of the participant;

[0025] Step S402, constructing a score database to store the emotion scores corresponding to different building appearance 3D model parameters, and evaluating the effect of the building appearance on alleviating emotional stress through statistical methods;

[0026] Step S403: identifying architectural features that induce positive emotions based on the emotion scores in the score database.

[0027] As a preferred technical solution, the architectural features that induce positive emotions include at least one of color matching, spatial layout, and light and shadow effects.

[0028] As a preferred technical solution, the emotion characteristic indicators include frequency domain characteristics, time domain characteristics, entropy value and complexity.

[0029] As a preferred technical solution, the step S5 includes:

[0030] Step S501, constructing a neural network model whose input is the parameters of the current building appearance 3D model and whose output is the influence of the parameters on the emotion regulation effect;

[0031] Step S502, adjusting parameters of the 3D model of the building appearance in order to minimize the difference between the emotion regulation effect predicted by the neural network model and the preset target;

[0032] Step S503, repeating step S502 until a preset termination condition is met.

[0033] As a preferred technical solution, the termination condition is:

[0034] After multiple consecutive iterations, the improvement in emotion regulation effect is less than the preset threshold.

[0035] Another aspect of the present invention provides an EEG feedback optimization system for building appearance visual features and emotion regulation, comprising:

[0036] The building appearance model construction module is used to construct a 3D building appearance model that matches the building characteristics of the target area and perform fine processing to obtain a 3D model library of multiple visual stimulus sources;

[0037] The AI ​​simulation environment module is used to construct a virtual scene including a 3D model of the building's exterior and at least one user, and configure the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations, and behavior patterns of the participants in the real environment;

[0038] EEG data acquisition and analysis module, used to obtain EEG signals of participants and extract emotional characteristic indicators;

[0039] The emotional impact assessment module is used to analyze the impact of the building appearance on emotional stress and positive emotions based on the predicted emotional state of participants under the current building appearance;

[0040] The feedback optimization mechanism module is used to predict the emotion regulation effect corresponding to the parameters based on the current building appearance 3D model parameters, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and iterate the parameters of the building appearance 3D model.

[0041] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0042] (1) Improving the objectivity and accuracy of the evaluation: The present invention uses signal processing algorithms and machine learning to directly analyze the changes in electrical signals in the cerebral cortex when receiving visual stimuli from different building appearances, thereby achieving an objective and accurate evaluation of the impact of building appearances on emotional states, reducing the impact of subjective bias and memory bias, and penetrating into the physiological level of emotion generation, significantly improving the scientificity and reliability of the evaluation.

[0043] (2) Enhance the real-time and interactive nature of the experiment: The AI ​​simulation environment module creates a highly immersive urban environment, supports real-time adjustment of viewing angles, scene layouts, and interaction methods, and provides participants with an experience similar to the real world. The real-time nature of the EEG data collection and analysis module enables the system to instantly provide feedback on changes in emotional states, providing an immediate basis for adjustment of architectural design.

[0044] (3) Personalization and optimization of design: Using machine learning of neural networks, the parameters of the building appearance model are automatically adjusted. After iterative calculation, the design scheme with the best emotion regulation effect is found, thus realizing personalized customization of the design. This function greatly improves the efficiency and accuracy of design optimization. At the same time, the support of visual interface tools enables designers to intuitively understand the optimization process and results, and quickly apply the optimization results to actual architectural design.

[0045] (4) Integrating cultural and environmental factors: The building appearance model construction module uses AI to combine local culture, climate, terrain and other characteristic factors to generate a 3D model of the building appearance with local characteristics, thereby enhancing the innovation and uniqueness of the architectural design. Through NLP analysis of local documents, historical records and resident interview data, the extracted cultural symbols, color preferences and architectural style elements make the design more in line with the needs and aesthetics of local residents, further enhancing the emotional experience and mental health level of residents. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of an EEG feedback optimization method for building appearance visual features and emotion regulation in an embodiment;

[0047] Figure 2 Schematic diagram of an EEG feedback optimization system for building appearance visual features and emotion regulation in an embodiment. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0049] Example 1

[0050] Regarding the problems existing in the above prior art, see Figure 1 This embodiment provides an EEG feedback optimization method for building appearance visual features and emotion regulation. By collecting and analyzing the electrical signals of the cerebral cortex, an objective, real-time and accurate evaluation of the impact of building appearance visual features on residents' emotions is achieved, and the building design is optimized accordingly to achieve a better emotion regulation effect. The method includes the following steps:

[0051] Step S1, building appearance model construction: Building a 3D building appearance model that matches the architectural features of the target area and performing fine processing to obtain a 3D model library of multiple visual stimulus sources.

[0052] Using AI technology, combined with local culture, climate, topography and other characteristic factors, we automatically generate 3D models of building exteriors with local characteristics. Through NLP analysis of local documents, historical records and resident interviews, we extract cultural symbols, color preferences and architectural style elements, and combine them with climate and topographic data to generate a preliminary design. In terms of model details, we achieve refined processing of color, shape and material to ensure the realism and layering of the model, and build a 3D model library containing multiple visual stimulus sources.

[0053] Step S1 may include the following sub-steps:

[0054] Step S101, data collection and preprocessing. Obtain one or more of local documents, historical records, and resident interview materials, and perform data cleaning, deduplication, and classification to extract cultural, color, and style information of the building's appearance.

[0055] Using NLP technology, key information is automatically extracted from local documents, historical records and resident interview materials, including cultural symbols, color preferences, architectural style elements, etc., and the collected data is pre-processed by cleaning, deduplication, classification and other operations to ensure the quality and accuracy of the data.

[0056] Step S102, characteristic factor integration: Analyze the impact of different climate factors and terrain factors on the building appearance, obtain the climate and terrain information of the building location, and integrate it with the cultural, color and style information of the building appearance to obtain a preliminary designed 3D model of the building appearance.

[0057] Combined with climate data such as temperature, humidity, sunshine duration and terrain information such as elevation, slope, and water distribution, AI algorithms are used to analyze the impact of these factors on the appearance of the building. The cultural, color, and style elements extracted by NLP are integrated with climate and terrain data to generate a preliminary design plan for the building appearance that conforms to local characteristics.

[0058] Step S103, fine modeling. Based on the preliminary designed 3D model of the building appearance, the color of the model is refined, and the material of the model is refined through 3D iterative rendering, taking into account the harmony and unity of the overall color tone and the effect of color changes under different materials and lighting conditions.

[0059] Based on the preliminary design, the color, shape and material of the model are refined. In terms of color, the harmonious unity of the overall tone and the changing effects of color under different materials and lighting conditions are considered; in shape design, traditional and modern design concepts are combined to create a recognizable architectural appearance; in terms of material, 3D rendering technology is used to simulate the light and shadow effects of real materials, and through iterative optimization algorithms, the model details are continuously adjusted to ensure the realism and layering of the model.

[0060] Preferably, step S103 also includes: adjusting model details.

[0061] Use the model detail adjustment tool to adjust the model. The model detail adjustment tool provides a color selector that allows users to select colors from preset color palettes or customize RGB / HSL values ​​to accurately match design requirements. At the same time, the integrated color editor supports users to adjust color saturation, brightness, contrast and other parameters to achieve fine-tuning of color effects. It also includes a wealth of shape design tools, such as line drawing, shape deformation, contour adjustment, etc. Users can edit and modify the model's contours, lines, surfaces, etc. as needed to create shapes that meet design requirements, and establish a comprehensive material library that includes a variety of real-world material effects (such as wood, stone, metal, etc.). At the same time, it supports users to customize materials and textures, and create unique material effects by uploading pictures or using graphic drawing tools to meet specific design needs.

[0062] Step S104: 3D model library construction and management: The refined building appearance 3D models are classified and stored in the 3D model library.

[0063] The refined building appearance models are stored in the 3D model library, which supports classification and retrieval by region, culture, style, etc., and provides model library update and maintenance functions to ensure the timeliness and integrity of the model library content.

[0064] Preferably, step S104 further includes: user interaction. Interaction with the user is achieved in a user interaction interface, in which an intuitive and user-friendly graphical interface is designed, allowing the user to input or select specific parameters such as culture, climate, terrain, etc. through controls such as drop-down menus, sliders, check boxes, etc. These parameters will serve as the basis for model construction and guide the subsequent design process. At the same time, a real-time preview function is integrated, so that when the user inputs or adjusts parameters, the user can immediately see the changes in the model effect on the interface. The window supports multi-angle viewing, zooming, rotation, etc., to help users evaluate the model more comprehensively.

[0065] Preferably, step S1 further includes:

[0066] Step S105: model evaluation and optimization feedback.

[0067] The integrated AI algorithm automatically evaluates the initially constructed model. The system identifies potential design problems or improvement points by analyzing the cultural symbols, color matching, shape design and other aspects of the model, and gives a corresponding evaluation report. According to the AI ​​evaluation results, it provides optimization suggestions or automatically adjusts the model parameters. By adjusting the color matching, shape design, material selection and other aspects, it improves the overall quality and conformity of the model. At the same time, it supports users to manually adjust the model parameters to further optimize the design effect, establishes a feedback loop mechanism, and allows users to make further manual adjustments or rebuild the model according to the evaluation results and feedback, and achieves the best model design effect through continuous iterative optimization.

[0068] Step S1 has the following characteristics:

[0069] (1) The automatic extraction of key information through NLP technology significantly improves the efficiency and accuracy of data collection. Compared with traditional manual collection methods, NLP technology can more comprehensively cover the information in local documents, historical records and resident interview materials. At the same time, through pre-processing operations such as data cleaning, deduplication, and classification, the quality and reliability of the data used for subsequent analysis are ensured. This improvement provides a solid foundation for the accurate extraction of cultural, color, style and other elements of architectural exterior design.

[0070] (2) In terms of characteristic factor fusion analysis, characteristic factor fusion analysis innovatively combines climate data and terrain information, and uses AI algorithms to analyze the impact of these factors on the appearance of buildings. This approach not only enhances the local characteristics of the design scheme, but also makes the appearance of the building more adapted to the local climate conditions and terrain environment. Compared with the design method that only considers single factors such as culture, color, and style, the fusion analysis strategy is more comprehensive and scientific, which helps to improve the practicality and aesthetics of architectural design.

[0071] (3) In terms of refined modeling, refined modeling was carried out on the basis of the preliminary design, and the color, shape and material of the model were deeply refined. In terms of color, the harmonious unity of the overall tone and the changing effect of color under different materials and lighting conditions were considered to make the appearance of the building more vivid and realistic. In terms of shape design, traditional and modern design concepts were combined to create an architectural appearance that is both recognizable and in line with the aesthetics of the times.

[0072] (4) In terms of materials, 3D rendering technology is used to simulate the light and shadow effects of real materials, and the model details are continuously adjusted through iterative optimization algorithms to ensure the realism and layering of the model. These refined processing measures have significantly improved the quality and visual effects of the building appearance model.

[0073] (5) In terms of 3D model library construction and management, the 3D model library construction and management provides designers with a convenient and efficient model library management tool. The model library supports classification and retrieval by region, culture, style, etc., allowing designers to quickly find architectural appearance models that meet their needs. The model library also provides update and maintenance functions to ensure the timeliness and integrity of the model library content. This improvement not only improves the work efficiency of designers, but also promotes the innovation and development of architectural appearance design.

[0074] (6) In terms of model detail adjustment, the model detail adjustment tools added in refined modeling greatly enhance the flexibility and freedom of users in the design process. The integration of the color selector tool and the color editor enables users to easily select colors from preset color palettes or achieve precise color matching through custom RGB / HSL values. At the same time, they can also adjust color saturation, brightness, contrast and other parameters to achieve ideal color effects. The rich shape design tools provide users with powerful editing capabilities. Whether it is line drawing, shape deformation or contour adjustment, they can easily cope with it, thereby creating unique shapes that meet design requirements. The establishment of a material library and the support for user-defined materials further broaden the range of material selection, allowing users to select or create real and unique material effects according to actual design needs.

[0075] (7) In terms of user interaction experience, the user interaction interface set up in the construction and management of the 3D model library greatly simplifies the user's operation process by designing an intuitive and user-friendly graphical interface. The introduction of controls such as drop-down menus, sliders, and check boxes allows users to easily input or select specific cultural, climate, terrain and other parameters, thereby guiding the subsequent model building process. The integration of real-time preview functions allows users to instantly see the changes in model effects when inputting or adjusting parameters, and supports multi-angle viewing, zooming, rotating and other operations to help users evaluate the model more comprehensively. This interactive method of instant feedback not only improves user work efficiency, but also enhances the user's design experience.

[0076] (8) In terms of model evaluation and optimization, the introduction of model evaluation and optimization feedback provides a systematic evaluation and optimization mechanism for model design. The integration of AI algorithms enables the system to automatically analyze the cultural symbols, color matching, shape design and other aspects of the model, identify potential design problems or improvement points, and give corresponding evaluation reports. Based on the AI ​​evaluation results, it will provide optimization suggestions or automatically adjust the model parameters to help users quickly improve the overall quality and compliance of the model. At the same time, it also supports the function of users manually adjusting model parameters, allowing users to make further manual adjustments or rebuild the model based on the evaluation results and feedback. The establishment of this feedback loop mechanism enables the model design to be continuously iterated and optimized, ultimately achieving the best design effect.

[0077] Step S2, AI simulates the environment. Construct a virtual scene including a 3D model of the building exterior and at least one user, and configure the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations and behavior patterns of the participants in the real environment.

[0078] Create a highly immersive urban environment, including streets, squares, parks and other scenes and natural elements. The system uses AI algorithms to analyze the movements, sight lines and behavior patterns of experimenters, and adjusts the perspective, scene layout and interaction methods in real time to provide an experience similar to the real world. At the same time, it supports multi-user online functions to improve the diversity and representativeness of the experiments.

[0079] Step S2 may include the following sub-steps:

[0080] Step S201, scene construction and dynamic adjustment. Obtain GIS data and real-world photos, construct a virtual scene including a 3D model of the building appearance and at least one user, adjust the perspective of the corresponding user in the virtual scene according to the actions, sight lines and behavior patterns of the participants in the real environment, dynamically adjust the scene layout according to the behavior patterns of the participants in the real environment, and predict and simulate the weather conditions and lighting changes in the virtual scene.

[0081] Using 3D modeling technology, combined with GIS data and real-world photos, a highly detailed and realistic urban environment is constructed. The buildings, streets, vegetation, etc. in the scene are carefully designed and optimized to ensure visual fidelity. AI algorithms are integrated to capture and analyze the movements, sight lines and behavior patterns of the experimenters in real time. According to the analysis results, the system can automatically adjust the viewing angle to ensure that the experimenters can observe the environment in the most comfortable way. At the same time, according to the behavior patterns of the experimenters, the scene layout is dynamically adjusted, such as adjusting the street crowd density, changing the park activity arrangements, etc., to simulate the dynamic changes of the real world. The introduction of natural element simulation technologies such as weather systems and lighting changes makes the virtual environment more vivid and realistic. For example, AI algorithms are used to predict and simulate weather conditions in different time periods (sunny, rainy, snowy, etc.), as well as lighting changes brought about by sunrise and sunset, to provide experimenters with a richer visual experience.

[0082] Step S202, multi-user online interaction and collaboration. The operation data of each participant is obtained in real time, and the user's behavior pattern and interaction intention between users are updated, and the behavior conflicts between users are detected and resolved.

[0083] It adopts a distributed computing architecture and efficient network communication protocols to support a large number of users online at the same time and interact in real time. The system can ensure that each user's operation can be responded to in a timely manner and accurately reflected in the virtual environment, thereby enhancing the collaborative effect between users. Through AI algorithms, the behavioral patterns and interaction intentions between multiple users are analyzed to achieve real-time synchronization of user behavior. When behavioral conflicts occur (such as two people trying to enter the same space at the same time), the system can automatically detect and resolve conflicts to ensure the stability and fairness of the virtual environment. It provides rich collaboration tools and an intuitive user interface, and supports real-time communication, task allocation, resource sharing and other functions between users. These tools can not only improve the efficiency of collaboration between users, but also promote communication and interaction between users, thereby enhancing the diversity and representativeness of the experiment.

[0084] Step S2 has the following characteristics:

[0085] (1) In terms of scene construction and dynamic adjustment, during the scene construction and dynamic adjustment process, a highly detailed and realistic urban environment is constructed by combining 3D modeling technology with GIS data and real-world photos. This combination not only ensures the visual fidelity of the scene, but also further enhances the user's sense of immersion by carefully designing and optimizing the elements such as buildings, streets, and vegetation in the scene. In addition, the integrated AI algorithm captures and analyzes the experimenter's movements, sight lines, and behavior patterns in real time, so that the perspective can be automatically adjusted to ensure that the experimenter can observe the environment in the most comfortable way. This personalized perspective adjustment greatly improves the user experience. At the same time, the scene layout can be dynamically adjusted according to the experimenter's behavior pattern, such as adjusting the street crowd density, changing the park activity arrangement, etc., to simulate the dynamic changes in the real world. This dynamic adjustment capability makes the virtual environment closer to reality and enhances the practicality and effectiveness of the experiment.

[0086] (2) In terms of introducing natural element simulation technologies such as weather systems and lighting changes, AI algorithms are used to predict and simulate weather conditions in different time periods (sunny, rainy, snowy, etc.) and lighting changes caused by sunrise and sunset, providing experimenters with a richer visual experience. This simulation of natural elements not only increases the realism of the virtual environment, but also makes the experimental environment more varied and complex, which helps experimenters conduct more comprehensive tests and analyses under different conditions.

[0087] (3) In terms of multi-user online interaction and collaboration, a distributed computing architecture and efficient network communication protocols are used in the multi-user online interaction and collaboration process to support a large number of users to be online at the same time and to interact in real time. This design ensures the stability and efficiency of the system, so that each user's operation can be responded to in a timely manner and accurately reflected in the virtual environment. At the same time, AI algorithms are used to analyze the behavior patterns and interaction intentions among multiple users, realize real-time synchronization of user behavior, effectively solve the problem of behavioral conflicts, and ensure the stability and fairness of the virtual environment. In addition, the rich collaboration tools and intuitive user interface provided not only improve the collaboration efficiency between users, but also promote communication and interaction between users, thereby enhancing the diversity and representativeness of the experiment. These technical features together constitute an efficient, stable, and realistic virtual experimental environment, providing experimenters with more comprehensive and in-depth testing and analysis methods.

[0088] Step S3, EEG data collection and analysis: Obtain the EEG signals of the participants and extract the emotional characteristic indicators.

[0089] The EEG signals of the experimenters are collected in real time, and the characteristic indicators related to emotions, such as α waves, β waves, and θ waves, are extracted using the wavelet transform signal processing algorithm. The emotional state recognition model is trained through the SVM machine learning algorithm to achieve real-time prediction of the emotional state of the experimenters.

[0090] Step S3 may include the following sub-steps:

[0091] Step S301, EEG data collection: The electrical signal of the EEG sensor is obtained, after analog-to-digital conversion and signal enhancement based on automatic gain control, the EGG signal of the participant is obtained after noise suppression and artifact removal.

[0092] High-precision, low-noise EEG sensors are designed as wearable or head-mounted devices that fit closely to the experimenter's scalp. These sensors can capture weak EEG signals and convert them into processable digital signals. The weak signals output by the sensors are enhanced through signal amplifiers, and filtering circuits are used to remove noise and interference, improve the signal-to-noise ratio of the signal, ensure that the EEG signals collected by all sensors can be recorded simultaneously and accurately, maintain the time synchronization of the signal, integrate automatic gain control systems, dynamically adjust the amplification factor according to the signal strength, avoid signal overload or distortion, and at the same time, regularly calibrate the sensors to ensure the accuracy and consistency of data collection. Use noise suppression and artifact removal algorithms to further purify EEG signals and reduce the impact of artifacts caused by eye movements, muscle activities, etc. on signal quality.

[0093] Step S302: signal processing and feature extraction: multi-scale decomposition is performed on the EGG signal to obtain a multi-band characteristic waveform, and a feature subset related to the emotional state is retained through feature screening as an emotional feature indicator.

[0094] The preprocessed EEG signal is decomposed into multiple scales using the wavelet transform algorithm to extract characteristic waveforms of different frequency bands, such as α wave, β wave, θ wave, etc. Wavelet transform can retain the local characteristics of the signal and is suitable for analyzing non-stationary EEG signals. Based on machine learning methods, the extracted features are screened and optimized to retain the feature subset most relevant to the emotional state. These features will be used as input to the emotion prediction model.

[0095] Preferably, the process of training the emotional state recognition model by using the SVM machine learning algorithm to achieve real-time prediction of the emotional state of the experimenter includes the following steps:

[0096] Step 1. Before the EEG signal is input into the SVM model, bandpass filter and notch filter technology are used to accurately remove noise sources such as power frequency interference (usually 50Hz or 60Hz), eye movement artifacts and electromyographic interference in the EEG signal to ensure signal quality. According to the needs of emotional state recognition, the continuous EEG signal is divided into segments of fixed length (such as 5 seconds to 30 seconds) for subsequent processing, and the emotional state label corresponding to each signal segment is marked. Z-score normalization is used to adjust the amplitude of the EEG signal to the same scale to ensure the stability of model training and the accuracy of prediction.

[0097] Step 2. Based on the use of wavelet transform to extract the frequency band characteristics of α waves, β waves, θ waves, etc., calculate the power spectral density and frequency band energy ratio of each frequency band to reflect the energy distribution characteristics of EEG signals in different frequency bands, extract statistical parameters such as the mean, variance, standard deviation, kurtosis, and skewness of the EEG signal, describe the basic characteristics of the signal in the time domain, and calculate the entropy value (such as sample entropy, approximate entropy) and complexity (such as Lempel-Ziv complexity) of the EEG signal to reveal the nonlinear dynamic behavior in the signal.

[0098] Step 3, to reduce feature redundancy and computational complexity, chi-square test or mutual information is used to evaluate the correlation between each feature and the emotional state, and irrelevant or weakly correlated features are eliminated. Recursive feature elimination technology is combined with the SVM model to gradually eliminate the features that contribute the least to the model performance. Model-based feature selection methods, such as gradient boosting machine or random forest, are used to automatically select important features during the model training process.

[0099] Step 4: In the model training stage, according to the EEG signal characteristics and the requirements of the emotion recognition task, the most suitable kernel function (such as the RBF kernel) is selected for training. A grid search or random search strategy is used, combined with cross-validation technology, to tune key parameters such as the penalty coefficient C of the SVM model and the γ parameter of the RBF kernel to find the optimal parameter combination.

[0100] Step 4, use an independent test data set to comprehensively evaluate the trained SVM model, measure the proportion of samples correctly classified by the model, evaluate the model's ability to recognize samples of specific emotional states, and comprehensively consider the harmonic mean of precision and recall to more comprehensively evaluate the model performance and display the classification of the model in each emotional state category, so as to facilitate the analysis of the causes of misclassification and missed classification.

[0101] Step 6, integrate the trained SVM model into the real-time prediction system, use high-precision EEG sensors to collect the EEG signals of the experimenters in real time, preprocess the collected signals, extract features and predict the emotional state. According to the prediction results, the system can give instant emotional state feedback or trigger corresponding control instructions to achieve real-time monitoring and effective intervention of the emotional state.

[0102] Step S3 has the following characteristics:

[0103] (1) In terms of EEG data collection, high-precision, low-noise EEG sensors are used and designed as wearable or head-mounted devices that fit closely to the scalp of the experimenter. This design effectively solves the limitations of traditional EEG equipment in signal capture, allowing the system to capture weak EEG signals more accurately. At the same time, through the integration of signal amplifiers and filtering circuits, the signal strength is significantly enhanced and noise and interference are removed, greatly improving the signal-to-noise ratio. In addition, it also has time synchronization to ensure that the EEG signals collected by all sensors can be recorded simultaneously and accurately, which is crucial for subsequent signal analysis and processing. The introduction of the automatic gain control system dynamically adjusts the amplification factor according to the signal strength, avoiding the problem of signal overload or distortion, and further improving the reliability of data collection. Regular calibration of the sensor also ensures the accuracy and consistency of data collection, providing a strong guarantee for the reliability of the experimental results. Finally, the noise suppression and artifact removal algorithms are used to further purify the EEG signal, reduce the impact of artifacts caused by eye movements, muscle activities, etc. on the signal quality, and make the collected signal purer, which is conducive to subsequent feature extraction and emotion prediction. Secondly,

[0104] (2) In terms of signal processing and feature extraction, the wavelet transform algorithm is used to perform multi-scale decomposition on the preprocessed EEG signal. This method not only retains the local characteristics of the signal, but is also particularly suitable for analyzing non-stationary EEG signals. Through wavelet transform, the system can extract characteristic waveforms of different frequency bands, such as α waves, β waves, θ waves, etc. These characteristic waveforms are closely related to different emotional states. Furthermore, the extracted features are screened and optimized based on machine learning methods, and the feature subsets most relevant to the emotional state are retained. This step effectively reduces the feature dimension and improves the effectiveness of the features, providing high-quality data support for the construction of the emotion prediction model. These technical features jointly realize efficient and accurate EEG data collection and analysis, and provide strong technical support for the objective evaluation and prediction of emotional states.

[0105] (3) In the data preprocessing stage, the use of bandpass filter and notch filter technology to remove noise sources such as power frequency interference, eye movement artifacts and electromyographic interference in the EEG signal is described in detail. This step ensures the signal quality of the input SVM model and lays a solid foundation for subsequent feature extraction and model training. In addition, the stability of model training and the accuracy of prediction are further improved by segmenting the continuous EEG signal into segments of fixed length and marking the emotional state labels, and using Z-score normalization to adjust the signal amplitude.

[0106] (4) In terms of feature extraction, not only the frequency band features such as α wave, β wave, and θ wave are extracted by wavelet transform, but also the parameters reflecting the energy distribution characteristics such as power spectrum density and frequency band energy ratio are calculated, as well as statistical parameters such as mean, variance, standard deviation, kurtosis, and skewness to describe the basic characteristics of the signal in the time domain. At the same time, nonlinear dynamic behavior characteristics such as entropy and complexity are introduced. These comprehensive feature extraction methods enable the model to capture more information related to the emotional state and improve the accuracy of emotion recognition.

[0107] (5) In the feature selection stage, through various means such as chi-square test, mutual information, recursive feature elimination and model-based feature selection methods, feature redundancy and computational complexity are effectively reduced, and the efficiency and performance of model training are improved. These feature selection techniques not only help to eliminate irrelevant or weakly relevant features, but also retain the feature subsets that are most relevant to the emotional state, providing more effective input data for the model. Fourth, in the model training stage, according to the characteristics of the EEG signal and the requirements of the emotion recognition task, the most suitable kernel function is selected for training, and the model parameters are tuned through grid search, random search and cross-validation technology to find the optimal parameter combination. This step ensures that the model can fully learn the rules in the data and improves the generalization ability and prediction accuracy of the model.

[0108] (6) In terms of real-time prediction system integration, the trained SVM model was integrated into the real-time prediction system, realizing the real-time acquisition, preprocessing, feature extraction and emotional state prediction of the experimenter's EEG signal. The system can give immediate emotional state feedback or trigger corresponding control instructions based on the prediction results, providing strong support for real-time monitoring and effective intervention of emotional states. This integrated solution not only improves the automation and practicality of the system, but also enables emotion recognition technology to be more widely used in practical scenarios.

[0109] Step S4, emotional impact assessment: Based on the predicted emotional state of the participants under the current building appearance, the impact of the building appearance on emotional stress and positive emotions is analyzed.

[0110] The emotional score is calculated according to the changes in characteristic indicators in the EEG data. By comparing and analyzing the differences in emotional scores under different building appearances, the effect of each building appearance on alleviating emotional stress is quantitatively evaluated. Special attention is paid to architectural features that can induce positive emotions, relaxation, and pleasure, and these features are used as the basis for optimization and adjustment.

[0111] Step S4 may include the following sub-steps:

[0112] Step S401, feature index mapping and emotion score calculation: The emotion feature index is mapped to a pre-built emotion score quantification system to obtain a score reflecting the emotional state of the participant.

[0113] The characteristic indicators obtained in the EEG data preprocessing and feature extraction stages, such as frequency domain characteristics (PSD, frequency band energy ratio), time domain characteristics (mean, variance, etc.), nonlinear characteristics (entropy value, complexity), etc., are mapped to the quantitative system of emotion scores. Through the preset emotion score calculation model (such as the mapping relationship based on the machine learning regression model), the characteristic indicators are converted into specific emotion scores to reflect the emotional state of the experimenters in numerical form.

[0114] Step S402: emotional score database and comparative analysis: Construct a score database to store the emotional scores corresponding to different building appearance 3D model parameters, and evaluate the effect of building appearance on alleviating emotional stress through statistical methods.

[0115] A database specifically used to store emotion score data is established to record the emotion scores of experimenters under different experimental conditions (such as different building appearances). The comparative analysis is responsible for extracting data from the database and conducting a comparative analysis of emotion scores across conditions (different building appearances). Statistical methods (such as t-test, analysis of variance, etc.) are used to quantitatively evaluate the effect of each building appearance on alleviating emotional stress.

[0116] Step S403, positive emotion feature recognition of the integrated recognition algorithm: Based on the emotion scores in the score database, architectural features that induce positive emotions are identified.

[0117] Identify architectural features that can induce positive emotions (such as relaxation and pleasure) from the emotion score data. These features may include specific color combinations, spatial layouts, light and shadow effects, etc.

[0118] Step S4 has the following characteristics:

[0119] (1) In terms of quantitative evaluation of emotional impact, through feature index mapping and emotion score calculation, the various feature indicators (including frequency domain features, time domain features and nonlinear features) obtained in the EEG data preprocessing and feature extraction stages are mapped to the quantitative system of emotion scores, realizing the numerical expression of emotional state. This quantitative evaluation method not only overcomes the subjectivity and ambiguity of traditional evaluation methods, but also improves the accuracy and objectivity of the evaluation. Through the preset emotion score calculation model, such as the mapping relationship based on the machine learning regression model, it can more accurately reflect the emotional state of the experimenter, providing reliable data support for the evaluation of emotional impact.

[0120] (2) In terms of cross-condition comparative analysis, the emotion score database and comparative analysis have achieved systematic storage and management of emotion score data. It also provides a cross-condition (such as different building appearance) emotion score comparative analysis function. By extracting data from the database and using statistical methods (such as t-test, variance analysis, etc.) for quantitative evaluation, it can intuitively show the effect of different building appearances on alleviating emotional stress. This comparative analysis not only helps to discover the intrinsic connection between architectural design and emotional state, but also provides a scientific basis for architectural design optimization.

[0121] (3) In terms of the recognition of positive emotional features, positive emotional feature recognition integrates advanced recognition algorithms and can automatically identify architectural features that can induce positive emotions (such as relaxation and pleasure) from the emotional score data. These features may include specific color matching, spatial layout, light and shadow effects, etc. They are of great significance for improving the comfort and livability of the building environment. By identifying these positive emotional features, more specific and targeted suggestions can be provided for architectural design, helping designers create more humane and comfortable architectural spaces.

[0122] Step S5, feedback optimization mechanism: Based on the current building appearance 3D model parameters, the emotion regulation effect corresponding to the parameters is predicted, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and the parameters of the building appearance 3D model are iterated.

[0123] The machine learning algorithm of the neural network is used to automatically adjust the parameters of the building appearance model, and the design scheme with the best emotion regulation effect is found through iterative calculation. At the same time, a visual interface tool is provided to display the optimization process and results, and support users to customize parameter settings and optimization target settings, so that designers can quickly apply the optimization results in actual building design.

[0124] Step S5 may include the following sub-steps:

[0125] Step S501, neural network model construction: A neural network model is constructed whose input is the parameters of the current building appearance 3D model and whose output is the influence of the parameters on the emotion regulation effect.

[0126] Design and build a neural network model suitable for the task of building appearance optimization. The model should be able to receive the current parameters of the building appearance as input and predict the impact of these parameters on the emotion regulation effect. The model architecture includes a convolutional neural network for processing image data and a fully connected layer for learning the complex relationship between parameters and emotion regulation effects.

[0127] Step S502: parameter adjustment and optimization: parameters of the building exterior 3D model are adjusted to minimize the difference between the emotion regulation effect predicted by the neural network model and the preset target.

[0128] Based on the prediction results of the neural network, a parameter adjustment strategy is implemented, which usually involves gradient descent or its variant algorithms, such as Adam, RMSprop, etc., to iteratively update the parameters of the building appearance model to minimize the difference between the predicted emotion regulation effect and the actual target. During the optimization process, the parameter constraints must also be considered, such as the availability of building materials, construction difficulty, etc.

[0129] Step S503, iterative calculation and convergence judgment. Step S502 is repeatedly executed until a preset termination condition is met.

[0130] Iterative calculation and convergence judgment Through multiple iterative calculations, the parameters of the building appearance model are gradually adjusted, and the emotion regulation effect after each adjustment is continuously evaluated. After each iteration, the preset convergence condition (such as the improvement of the emotion regulation effect after multiple consecutive iterations is less than a certain threshold) is used to judge whether the optimal solution is reached or close to the local optimal solution.

[0131] Step S5 has the following characteristics:

[0132] (1) In terms of intelligent optimization, in the construction of the neural network model, by designing and building a neural network model suitable for the task of building appearance optimization, the intelligent processing of building appearance parameters is realized. The model can not only receive the current parameters of the building appearance as input, but also predict the impact of these parameters on the emotion regulation effect, thus providing a scientific basis for the optimization process. In particular, the introduction of the convolutional neural network in the model architecture enables the model to efficiently process image data and capture the key features of the building appearance; the setting of the fully connected layer helps to learn the complex relationship between the parameters and the emotion regulation effect, thereby improving the accuracy of the prediction.

[0133] (2) In terms of parameter adjustment and optimization, parameter adjustment and optimization are based on the prediction results of the neural network and implement parameter adjustment strategies. By adopting gradient descent or its variant algorithms (such as Adam, RMSprop, etc.), the parameters of the building appearance model can be iteratively updated to minimize the difference between the predicted emotion regulation effect and the actual target. This parameter adjustment method not only improves the efficiency of the optimization process, but also ensures the accuracy of the optimization results. At the same time, the optimization process also fully considers the parameter constraints, such as the availability of building materials, construction difficulty, etc., so that the optimization results are more in line with the actual situation.

[0134] (3) In terms of the convergence judgment mechanism, iterative calculation and convergence judgment gradually adjust the parameters of the building appearance model through multiple iterative calculations, and continuously evaluate the emotion regulation effect after each adjustment. After each iteration, the preset convergence condition (such as the improvement of the emotion regulation effect after multiple consecutive iterations is less than a certain threshold) is used to judge whether the optimal solution is reached or close to the local optimal solution. This convergence judgment mechanism not only ensures the stability and reliability of the optimization process, but also avoids unnecessary waste of computing resources.

[0135] Example 2

[0136] Based on Example 1, see Figure 2 , this embodiment provides an EEG feedback optimization system for building appearance visual features and emotion regulation, including:

[0137] (1) Building appearance model construction module, which is used to construct a 3D building appearance model that matches the architectural characteristics of the target area and perform fine processing to obtain a 3D model library of multiple visual stimulus sources.

[0138] Preferably, the building appearance model construction module specifically includes: a data collection and preprocessing submodule, a characteristic factor fusion analysis submodule, a refined modeling submodule and a 3D model library construction and management submodule.

[0139] Preferably, a model detail adjustment tool module is also provided in the refined modeling submodule.

[0140] Preferably, a user interaction interface module is also provided in the 3D model library construction and management submodule.

[0141] Preferably, a model evaluation and optimization feedback module may also be provided in the building appearance model construction module.

[0142] (2) An AI simulation environment module is used to construct a virtual scene including a 3D model of the building exterior and at least one user, and configure the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations, and behavior patterns of the participants in the real environment.

[0143] Preferably, the AI ​​simulation environment module includes a scene construction and dynamic adjustment sub-module and a multi-user online interaction and collaboration sub-module.

[0144] (3) EEG data acquisition and analysis module, used to obtain the EEG signals of participants and extract emotional characteristic indicators.

[0145] Preferably, the EEG data acquisition and analysis module includes an EEG data acquisition subsystem and a signal processing and feature extraction subsystem.

[0146] (4) Emotional impact assessment module, which is used to analyze the impact of building appearance on emotional stress and positive emotions based on the predicted emotional state of participants under the current building appearance.

[0147] The emotion impact assessment module includes a feature index mapping and emotion score calculation unit, an emotion score database and comparative analysis module, and a positive emotion feature recognition module.

[0148] (5) A feedback optimization mechanism module is used to predict the emotion regulation effect corresponding to the parameters based on the current building appearance 3D model parameters, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and iterate the parameters of the building appearance 3D model.

[0149] The feedback optimization mechanism module includes a neural network model building unit, a parameter adjustment and optimization module, and an iterative calculation and convergence judgment module.

[0150] This system integrates multiple modules, aiming to explore and optimize the effect of building appearance on emotion regulation by combining AI technology, EEG data analysis and emotion assessment, and neural network feedback optimization mechanism. It has the following characteristics:

[0151] (1) The building appearance model construction module uses advanced AI technology to deeply explore and analyze local culture, climate, topography and other characteristic factors. It uses NLP technology to process a large amount of local documents, historical records and resident interview data to accurately extract cultural symbols, color preferences and architectural style elements. Combined with climate and topographic data, this module can automatically generate a 3D model of the building appearance with distinct local characteristics, and perform fine processing on the color, shape and material of the model details to ensure the realism and layering of the model. Finally, it constructs a 3D model library containing rich visual stimulation sources.

[0152] (2) The AI ​​simulation environment module uses highly immersive virtual technology to create a realistic urban environment, including streets, squares, parks and other diverse scenes and natural elements. Through AI algorithms, the movements, sight lines and behavior patterns of experimenters are analyzed in real time. The system can intelligently adjust the perspective, scene layout and interaction mode to provide experimenters with an experience that is extremely similar to the real world. At the same time, the module supports multi-user online functions, which effectively improves the diversity and representativeness of the experiment and makes the evaluation results more comprehensive and accurate.

[0153] (3) In the EEG data acquisition and analysis module, the system collects the EEG signals of the experimenters in a simulated environment in real time, and uses advanced signal processing algorithms such as wavelet transform to accurately extract characteristic indicators closely related to emotions, such as α waves, β waves, θ waves, etc. Subsequently, the emotion state recognition model is trained through machine learning algorithms such as SVM to achieve real-time prediction and accurate analysis of the emotional state of the experimenters.

[0154] (4) The emotional impact assessment module is based on the changes in characteristic indicators in the EEG data. Through the preset emotional score calculation model, the characteristic indicators are converted into specific emotional scores, which intuitively reflect the emotional state of the experimenters in numerical form. By comparing and analyzing the differences in emotional scores under different building appearances, this module can quantitatively evaluate the effect of each building appearance on alleviating emotional stress, and pay special attention to those architectural features that can induce positive emotions (such as relaxation and pleasure). These features will serve as an important basis for subsequent optimization and adjustment.

[0155] (5) The feedback optimization mechanism module uses a neural network machine learning algorithm to automatically adjust the parameters of the building appearance model according to the results of the emotion impact assessment. Through iterative calculation and continuous iteration of the optimization algorithm, the system can gradually find the design solution with the best emotion regulation effect. At the same time, the module also provides a visual interface tool to intuitively display the optimization process and results, support user-defined parameter settings and optimization target settings, and facilitate designers to quickly apply the optimization results to actual building design. The whole process forms a closed-loop feedback optimization system to continuously iterate and improve the effect of the building appearance on emotion regulation.

[0156] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An EEG feedback optimization method for building appearance visual features and emotion regulation, characterized in that: The steps include: Step S1, constructing a 3D model of the building appearance that matches the building characteristics of the target area and performing fine processing to obtain a 3D model library of multiple visual stimulus sources; Step S2, constructing a virtual scene including a 3D model of the building appearance and at least one user, and configuring the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations and behavior patterns of the participants in the real environment; Step S3, obtaining the EEG signal of the participant and extracting the emotion characteristic index; Step S4, analyzing the influence of the building appearance on emotional stress and positive emotions based on the predicted emotional state of the participants under the current building appearance; Step S5, based on the current building appearance 3D model parameters, predict the emotion regulation effect corresponding to the parameters, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and iterate the parameters of the building appearance 3D model.

2. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1, characterized in that: The step S1 comprises: Step S101, obtaining one or more of local documents, historical records, and resident interview data, and performing data cleaning, deduplication, and classification processing to extract cultural, color, and style information of the building appearance; Step S102, analyzing the influence of different climate factors and terrain factors on the building appearance, obtaining the climate and terrain information of the building location, and integrating it with the culture, color and style information of the building appearance to obtain a preliminary designed 3D model of the building appearance; Step S103, based on the preliminary designed 3D model of the building appearance, the color of the model is refined by considering the harmony and unity of the overall color tone and the effect of color changes under different materials and lighting conditions, and the material of the model is refined by 3D iterative rendering; Step S104: Classify and store the refined building appearance 3D models in a 3D model library.

3. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1 is characterized in that: The step S2 comprises: Step S201, obtaining GIS data and real-world photos, constructing a virtual scene including a 3D model of the building appearance and at least one user, adjusting the perspective of the corresponding user in the virtual scene according to the actions, sight lines and behavior patterns of the participants in the real environment, dynamically adjusting the scene layout according to the behavior patterns of the participants in the real environment, and predicting and simulating weather conditions and lighting changes in the virtual scene; Step S202, obtaining the operation data of each participant in real time, updating the user's behavior pattern and the interaction intention between users, and detecting and resolving the behavior conflicts between users.

4. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1, characterized in that: The step S3 comprises: Step S301, obtaining the electrical signal of the EEG sensor, performing analog-to-digital conversion and signal enhancement based on automatic gain control, and obtaining the EGG signal of the participant through noise suppression and artifact removal; Step S302 , performing multi-scale decomposition on the EGG signal to obtain a multi-band characteristic waveform, and retaining a feature subset related to the emotional state as an emotional feature indicator through feature screening.

5. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1, characterized in that: The step S4 comprises: Step S401, mapping the emotion characteristic index to a pre-constructed emotion score quantification system to obtain a score reflecting the emotional state of the participant; Step S402, constructing a score database to store the emotion scores corresponding to different building appearance 3D model parameters, and evaluating the effect of the building appearance on alleviating emotional stress through statistical methods; Step S403: identifying architectural features that induce positive emotions based on the emotion scores in the score database.

6. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 5, characterized in that: The architectural features that induce positive emotions include at least one of color matching, spatial layout, and light and shadow effects.

7. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1, characterized in that: The emotion characteristic indicators include frequency domain characteristics, time domain characteristics, entropy value and complexity.

8. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 1, characterized in that: The step S5 comprises: Step S501, constructing a neural network model whose input is the parameters of the current building appearance 3D model and whose output is the influence of the parameters on the emotion regulation effect; Step S502, adjusting parameters of the 3D model of the building appearance in order to minimize the difference between the emotion regulation effect predicted by the neural network model and the preset target; Step S503, repeating step S502 until a preset termination condition is met.

9. The EEG feedback optimization method for building appearance visual features and emotion regulation according to claim 8, characterized in that: The termination conditions are: After multiple consecutive iterations, the improvement in emotion regulation effect is less than the preset threshold.

10. An EEG feedback optimization system for building appearance visual features and emotion regulation, characterized in that: include: The building appearance model construction module is used to construct a 3D building appearance model that matches the building characteristics of the target area and perform fine processing to obtain a 3D model library of multiple visual stimulus sources; The AI ​​simulation environment module is used to construct a virtual scene including a 3D model of the building's exterior and at least one user, and configure the perspective and layout of the corresponding user in the virtual scene according to the actions, realizations, and behavior patterns of the participants in the real environment; EEG data acquisition and analysis module, used to obtain EEG signals of participants and extract emotional characteristic indicators; The emotional impact assessment module is used to analyze the impact of the building appearance on emotional stress and positive emotions based on the predicted emotional state of participants under the current building appearance; The feedback optimization mechanism module is used to predict the emotion regulation effect corresponding to the parameters based on the current building appearance 3D model parameters, so as to minimize the difference between the predicted emotion regulation effect and the preset target, and iterate the parameters of the building appearance 3D model.

Images