Indoor decoration intelligent design method and system based on large model

Abstract

The invention relates to the technical field of indoor design and intelligent buildings, and discloses an indoor decoration intelligent design method and system based on a large model, and the method comprises the steps: collecting user activity data, constructing a user behavior model, and generating a personalized activity mode library; based on the user behavior data, a time-varying Markov decision process is applied to construct space function demand models of different time periods, and a time-sequenced space scoring function is generated; analyzing a cognitive processing process of a user in a multi-task environment through a cognitive scientific modeling method, and calculating a cognitive load index of a spatial layout scheme; applying an improved ant colony algorithm to generate an optimal space layout scheme meeting multiple targets; and optimizing the layout model based on user feedback and monitoring data, and generating a decoration implementation scheme. According to the intelligent design method, the user behavior pattern, the biological rhythm and the cognitive load can be fully considered, and the space use efficiency and the function adaptability are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of interior design and intelligent building technology, and more particularly to a method and system for intelligent design of interior decoration based on a large model. Background Art

[0002] With the diversification of modern living needs, interior spaces need to support multiple functions and tasks simultaneously. Traditional interior design primarily focuses on static aesthetics, often based on designers' experience and simple user requirements. It lacks quantitative analysis and consideration of the human body's dynamic behavioral needs in daily life, changes in biological rhythms, and the cognitive load associated with multitasking. This results in renovated spaces that, while aesthetically pleasing, often suffer from poor circulation, low space utilization, and poor functional adaptability during actual use. This is particularly evident in different time periods and multitasking scenarios.

[0003] Existing solutions often rely on static planning methods, such as CAD design software and 3D modeling tools. While these tools can provide basic spatial planning and aesthetic design, they fail to effectively account for the dynamic nature of user behavior and cognitive load. Some smart home solutions have initially introduced user behavior analysis, but they still cannot accurately model user activity trajectories. They also lack comprehensive consideration of biorhythm cycles and cognitive processing mechanisms, making them unable to provide users with all-weather, low-cognitive-load smart space solutions.

[0004] In the existing technology, although it is possible to generate multiple layout schemes based on simple user needs and take into account the requirements of biological rhythms for the light environment, it does not take into account the dynamic and time-varying needs of user behavior. At the same time, it is limited to the lighting system and has not been extended to the overall space layout design. There is also a lack of analysis and optimization design of the actual user behavior.

[0005] Overall, existing technologies lack an intelligent interior design approach that fully integrates user dynamic behavior patterns, circadian rhythmic changes, and cognitive load assessment. This makes it difficult to achieve precise, personalized, and dynamic configuration of limited spaces, and fails to meet modern users' growing demand for multifunctional and efficient spaces. Therefore, a new intelligent interior design approach is urgently needed that can combine advanced artificial intelligence technologies with a deep understanding of user behavior and cognitive characteristics to provide optimal spatial layout solutions that truly adapt to users' all-day needs. Summary of the Invention

[0006] The present invention provides a large-model-based intelligent design method and system for interior decoration, which solves the technical problem in related technologies of lacking comprehensive consideration of users' dynamic behavior patterns, biological rhythms and cognitive load.

[0007] A first aspect of the present invention provides an intelligent design method for interior decoration based on a large model, comprising the following steps: Collect user activity data, build user behavior models, and generate a personalized activity pattern library; Based on the personalized activity pattern library, a time-varying Markov decision process is applied to construct spatial functional demand models for different time periods and generate a time-series spatial scoring function. Based on the temporal spatial scoring function, cognitive science modeling methods are used to analyze the user's cognitive processing in a multi-tasking environment and calculate the cognitive load index under different spatial layout schemes. According to the spatial scoring function and cognitive load index, an improved ant colony algorithm is applied to generate the optimal spatial layout solution that meets multiple objectives. Based on user feedback on the optimal space layout plan and continuously monitored behavioral data, the space layout model is optimized and a detailed decoration implementation plan is generated.

[0008] Furthermore, the collection of user activity data includes: collecting the user's location movement trajectory, activity status and physiological parameters through smart wearable devices, and monitoring the user's specific location, stay duration and activity type indoors through an indoor sensor network.

[0009] Furthermore, the step of constructing the user behavior model includes: Using the improved long short-term memory network algorithm, the collected raw data are subjected to feature extraction and pattern recognition; Divide user activities into multiple categories and record the spatiotemporal distribution characteristics, frequency characteristics, and correlation characteristics of each category of activities; Extract activity patterns within a 24-hour period, including peak and trough times for different activities.

[0010] Furthermore, the steps of constructing spatial functional demand models for different time periods using the time-varying Markov decision process include: Construct an individual biorhythm model and define biorhythm influencing factors; Divide the indoor space into multiple functional areas and build a probability model of users' spatial state transitions at different times; Based on the biological rhythm model and state transition probability, the functional demand scores of each spatial area at different time periods are calculated.

[0011] Furthermore, the steps of analyzing the user's cognitive processing in a multi-tasking environment by the cognitive science modeling method include: Apply task analysis techniques to decompose common activities in home settings into basic task units and identify the types and amounts of cognitive resources required for each task unit; Based on the ACT-R cognitive architecture, a user's cognitive processing model in a multi-tasking environment is constructed; Calculate the switching cost function between different tasks; Based on the task switching cost and task execution probability, the comprehensive cognitive load index under a specific spatial layout scheme is calculated.

[0012] Furthermore, the step of applying the improved ant colony algorithm to generate an optimal spatial layout solution that meets multiple objectives includes: The indoor space is divided into grid units and the spatial layout is represented by parametric coding. Construct a multi-objective optimization function that comprehensively considers spatial utility, cognitive adaptability, aesthetic value, and construction cost objectives; An improved ant colony algorithm that introduces the Pareto optimization principle and adaptive pheromone update strategy is used to solve multi-objective optimization problems. Develop an interactive optimization interface that enables users to browse and interactively adjust the set of non-dominated solutions recommended by the system.

[0013] Furthermore, the multi-objective optimization function includes: Spatial utility function, which measures the efficiency of the layout in space utilization; Cognitive load function, which evaluates the impact of layout on user cognitive load; Aesthetic value function, which evaluates the aesthetic value of the layout; A construction cost function that evaluates the renovation costs required to realize the layout.

[0014] Furthermore, the steps of optimizing the space layout model and generating a detailed decoration implementation plan include: Build a user feedback collection and processing mechanism to integrate users' subjective evaluation and preferences of layout solutions into the optimization model; Build a time-series virtual usage scenario simulator to verify the dynamic adaptability of the optimized layout in daily use; Adopting incremental adjustment strategies and reinforcement learning frameworks to iteratively optimize layout solutions; Generate detailed decoration implementation guidance documents and 3D visual rendering effects.

[0015] Furthermore, the improved ant colony algorithm includes the following steps: Calculating multiple objective function values ​​for each layout solution; Identify non-dominated solutions based on the Pareto dominance relationship and update the set of non-dominated solutions; Update pheromone for each target according to the objective function value; Dynamically adjust the target weights based on the performance distribution of the ant population on each target; All solutions are sorted using the fast non-dominated sorting algorithm and the Pareto front solution set is updated.

[0016] A second aspect of the present invention provides a large-model-based intelligent interior decoration design system, which is used to execute the above-mentioned large-model-based intelligent interior decoration design method, including: Data collection module, used to collect user activity data, build user behavior models, and generate personalized activity pattern libraries; The demand mapping module is used to construct spatial functional demand models for different time periods based on user behavior data and apply the time-varying Markov decision process to generate a time-series spatial scoring function; The cognitive assessment module is used to analyze the user's cognitive processing in a multi-tasking environment through cognitive science modeling methods and calculate the cognitive load index under different spatial layout schemes; An optimization generation module is used to apply an improved ant colony algorithm to generate an optimal spatial layout solution that meets multiple objectives based on the results of biorhythm space demand and cognitive load assessment; Implementation guidance module, which is used to optimize the space layout model and generate detailed decoration implementation plans based on user feedback and continuously monitored behavioral data The beneficial effects of the present invention are as follows: by integrating multi-source data collection, user behavior modeling, biorhythm analysis, cognitive load assessment, and multi-objective optimization algorithms, it solves the technical problem of traditional interior design's inability to accurately perceive and adapt to users' dynamic behavioral needs. Compared with traditional design methods, the spatial layout design using the present method improves space utilization efficiency and reduces traffic flow conflicts, especially in small spaces with limited area. By introducing biorhythm analysis and a time-varying Markov decision process, the present method can accurately predict spatial functional requirements at different time periods, thereby improving user satisfaction in the optimized space at different time periods. Through the learning ability of large-scale model technology, it can accurately capture individual user characteristics and preferences and generate highly personalized decoration plans accordingly. The system can automatically adjust optimization strategies and parameter weights according to the specific needs of different user groups, providing tailored solutions. Through multi-objective optimization algorithms and Pareto front analysis, the present method solves the problem of insufficient functionality in traditional decoration configurations while ensuring visual aesthetics. Through precise space planning and functional analysis, it can reduce unnecessary decoration elements and material waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the interior decoration intelligent design method based on a large model of the present invention. DETAILED DESCRIPTION

[0018] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses an intelligent design method for interior decoration based on a large model, such as Figure 1 As shown, the following steps are included: Step 1: Collect user activity data, build a user behavior model, and generate a personalized activity pattern library.

[0020] It includes the following sub-steps: Step 1.1: Data acquisition system construction; Build a multi-source data collection system consisting of smart wearable devices and an indoor sensor network. It should be understood that the smart wearable devices primarily collect user location, movement, activity status, and physiological parameters (such as heart rate and step count). The indoor sensor network, consisting of infrared motion sensors, pressure sensors, and cameras installed at various locations indoors, monitors the user's location, duration, and activity type. Furthermore, the system provides continuous user behavior data while protecting user privacy.

[0021] Step 1.2: Extract time series data features; Leveraging the time series data processing capabilities of the large model, we perform feature extraction and pattern recognition on the collected raw data. Specifically, we use an improved Long Short-Term Memory (LSTM) algorithm, which, by introducing a temporal attention mechanism, can effectively capture long-term dependencies and periodic patterns in user behavior data. It should be noted that the mathematical expression of the algorithm is: ; in, Represents the hidden state at time t, representing the model's internal representation of the user's behavior at the current moment; The input data at time t (including location, activity type, duration, etc.) is the raw data collected by the sensor; Represents the hidden state of the previous moment, which contains historical behavior information; Represents the unit state at the previous moment and stores long-term memory information; Represents the time attention weight, which is used to emphasize the historical time points that are more important for the current prediction; Indicates the current time point, Represents the long short-term memory network function.

[0022] The specific implementation of the improved LSTM algorithm includes the following steps: Input gate calculation: controls the degree to which new input information flows into the cell state: ; in, Represents the output value of the input gate, which controls the input level of new information; Represents the sigmoid activation function, which maps the input to between 0 and 1 and is used to control the flow of information; Represents the weight matrix of the input gate, which determines which new information needs to be saved; Indicates that the vector and Concatenate into a new vector; Represents the bias term of the input gate, which is used to adjust the activation threshold; Indicates the current time point, The initials for "Input" identify parameters and variables related to the input gate.

[0023] Forget gate calculation: controls the degree of retention of the unit state at the previous moment: ; in, Represents the output value of the forget gate, which controls the degree of retention of historical information; Represents the sigmoid activation function, which maps the input to between 0 and 1 and is used to control the flow of information; Represents the weight matrix of the forget gate, which determines which old information needs to be discarded; Indicates that the vector and Concatenate into a new vector; Represents the bias term of the forget gate, which is used to adjust the activation threshold.

[0024] Unit state update: combining new information with forgotten historical information: ; ; in, Represents the candidate unit state, including new information that may be updated; represents the hyperbolic tangent activation function, which maps values ​​to between -1 and 1; A weight matrix representing the cell state, used to generate new candidate values; Indicates that the vector and Concatenate into a new vector; A bias term representing the cell state, used to adjust the candidate value; Represents the unit state at the current moment and stores long-term memory information; Represents the output value of the forget gate, which controls the degree of retention of historical information; Represents the unit state at the previous moment and stores long-term memory information; Represents the output value of the input gate, which controls the degree of input of new information.

[0025] Output gate calculation: controls the degree to which the cell state affects the current output: ; in, Represents the output value of the output gate, which controls the degree of output of the unit state to the hidden state; Represents the sigmoid activation function, which maps the input to between 0 and 1 and is used to control the flow of information; The weight matrix representing the output gate, which controls which parts of the cell state will be output; Indicates that the vector and Concatenate into a new vector; Represents the bias term of the output gate, which is used to adjust the activation threshold.

[0026] Output calculation: Generate the output at the current moment: ; in, represents the output at time t, which represents the model's internal representation of the user's behavior at the current moment; Represents the output value of the output gate, which controls the degree of output of the unit state to the hidden state; represents the hyperbolic tangent activation function, which maps values ​​to between -1 and 1; Indicates the unit state at the current moment and stores long-term memory information.

[0027] The temporal attention mechanism is implemented as follows: For the historical hidden state sequence Calculate the attention score: ; in, represents the attention score of historical moment j, reflecting its importance to the current prediction; is a learnable parameter vector used to calculate the attention score; Represents a vector The transpose of represents the hyperbolic tangent activation function, which maps values ​​to between -1 and 1; is a learnable parameter matrix used to transform historical hidden states; represents the hidden state at historical moment j; is a learnable parameter matrix used to transform the current hidden state; Represents the hidden state of the previous moment, which contains historical behavior information. 、 、 Represent the hidden states at time 1, time 2, and time t-1 respectively, An index representing a historical moment.

[0028] Normalize the attention scores to generate attention weights: ; in, represents the normalized attention weights, ensuring that the sum of all weights is 1; represents the natural exponential function; Representing historical moments The attention score of , which reflects its importance to the current prediction; represents the sum of the index scores for all values ​​of i from 1 to t-1; An index representing a specific historical moment; Indicates the previous moment, An index representing a historical moment.

[0029] Generate context vector based on attention weights: ; in, represents the context vector, which is the weighted sum of historical hidden states; It means summing all j values ​​from 1 to t-1; represents the normalized attention weights, ensuring that the sum of all weights is 1; represents the hidden state at historical moment j; An index representing a historical moment; Indicates the previous moment.

[0030] Combine the context vector with the current input: ; in, Represents the enhanced input, which combines the original input and historical context information; Indicates that the vector and Concatenate into a new vector; Represents the input data at time t; represents the context vector, which is the weighted sum of historical hidden states.

[0031] Update the LSTM using augmented input: ; in, represents the updated hidden state at time t, which represents the model's internal representation of the user's behavior at the current moment; represents the long short-term memory network function; Represents the enhanced input, which combines the original input and historical context information; Represents the hidden state of the previous moment, which contains historical behavior information; Represents the unit state at the previous moment and stores long-term memory information.

[0032] In specific application scenarios, this improved LSTM algorithm effectively identifies patterns in regular user activity by capturing long-term dependencies within user behavior sequences. For example, in a small apartment scenario, by analyzing 30 consecutive days of user location and activity data, the system can identify a fixed weekday morning activity sequence: washing up, making breakfast, and leaving the house, and capture the temporal and spatial relationships between these activities. The temporal attention mechanism automatically identifies key moments in a sequence. For example, when predicting a user's breakfast activity, the system focuses more on the previous night's shopping behavior and the morning's wake-up time, rather than the previous day's work status, thereby improving prediction accuracy.

[0033] Step 1.3: Build a user behavior pattern library; Based on the feature extraction results, a user behavior pattern library is constructed. This library categorizes user activities into multiple categories (such as work, rest, dining, and socializing), and records the spatiotemporal distribution, frequency, and correlation characteristics of each activity type. Specifically, the system identifies the action space requirements for different activity types, namely the minimum spatial dimensions and optimal spatial configuration required to complete a specific activity. Furthermore, the system extracts activity patterns within a 24-hour cycle, including peak and trough times for different activities, providing foundational data for subsequent biorhythm analysis.

[0034] Step 2: Based on the personalized activity pattern library, the time-varying Markov decision process is applied to construct the spatial functional demand model for different time periods and generate a time-series spatial scoring function.

[0035] It includes the following sub-steps: Step 2.1, biorhythm modeling; Based on user activity data, an individual biorhythm model is constructed. This model identifies the user's individual biorhythm characteristics by analyzing the user's activity patterns, physiological parameter changes, and attention level fluctuations within a 24-hour cycle. It should be noted that the biorhythm influencing factors It is defined as the user's physiological readiness and preference for activity type i at time t, and its mathematical expression is: ; in, represents the user's biorhythm impact factor on activity type i at time t, quantifying the degree of influence of biorhythms on specific activities; Indicates the user's physiological adaptability to activity i at time t (such as heart rate, muscle fatigue, etc.), reflecting the degree to which the physical condition supports the activity; It represents the adaptability of the user's emotional state to activity i at time t, reflecting the impact of emotions on activities; It represents the user's cognitive adaptability to activity i at time t (such as attention level, decision-making efficiency, etc.), reflecting the degree of support for the activity by the cognitive state; The weight coefficient representing physiological adaptability determines the importance of physiological factors; The weight coefficient representing the adaptability of the emotional state determines the importance of the emotional factor; The weight coefficient representing the adaptability of cognitive ability determines the importance of cognitive factors; Indicates a point in time; An index representing the activity type.

[0036] Step 2.2: Modeling the time-varying Markov decision process; A time-varying Markov decision process (TV-MDP) is used to construct a probability model for users' spatial state transitions at different time periods. This model divides the indoor space into multiple functional areas (states). The user's movement between areas is considered a state transition, and the transition probability will change at different time periods. Therefore, the state transition probability matrix Defined as: ; in, represents the state transition probability matrix at time t, describing the probability distribution of users moving between different spatial regions; Indicates that the user at time t is from the region Move to area The probability of ,reflects the spatial migration pattern of users at a specific time point; Indicates a point in time; Indicates the index of the starting area; Indicates the index of the target area; It should be noted that these probabilities will be dynamically adjusted according to the change of time t, reflecting the changes in users' space usage preferences at different time periods.

[0037] The specific implementation of the time-varying Markov decision process (TV-MDP) includes the following steps: State space definition: Divide the indoor space into N functional areas, each area corresponds to a state , the state space is recorded as: ; in, Indicates the The status corresponding to each functional area; Represents the region index; Indicates the total number of functional areas; Represents the state space, which contains all possible states; represents the set of integers from 1 to m; 、 、 Respectively represent each state in the state space; Symbols that represent belonging to a set.

[0038] Time window division: Divide 24 hours into M time windows, such as one window per hour, for a total of 24 windows, recorded as: ; in, Represents a time window set, including all divided time windows; Indicates a specific time window; Indicates the total number of time windows, that is, the number of windows divided into 24 hours a day (for example, if there is one window per hour, then M=24); 、 、 Represent the time windows of the 1st, 2nd, and Mth moments respectively.

[0039] Data collection and preprocessing: Collect user location sequence data in different time windows, and perform outlier processing and data smoothing. Time-varying transition probability matrix calculation: For each time window , calculate the conditional transition probability: ; in, Indicates that in a specific time window Within, users from the area Move to area The conditional probability of Indicates that in a specific time window Within, users from the area Move to area The number of times is the frequency obtained from the observation data; Indicates that in a specific time window Within, users from the area The total number of moves to any area is used to normalize the probability value; Indicates the starting area; Indicates the target area; represents any possible target area; Indicates the total number of functional areas; Represents the region index; Indicates a specific time window; Indicates state transition; Represents the conditional symbol in conditional probability; Indicates the number of All Sum of values, Indicates the total number of time windows, that is, the number of windows divided into 24 hours a day (for example, if there is one window per hour, then M=24); 、 、 Represent the time windows of the 1st, 2nd, and Mth moments respectively.

[0040] Time-varying reward function definition: For each state-time pair , define the reward function based on the user's stay time, activity type and comfort feedback in this state and time in, Indicates that in the time window In state Instant rewards based on user satisfaction with activities in the area; Indicates the status of a specific functional area; Indicates a specific time window; Represents a state-time pair, which refers to a specific state within a specific time window.

[0041] Time-varying value function calculation: Use the time-varying Bellman equation to calculate the value of each state-time pair; ; in, Indicates that in the current time window In state The value of reflects the importance of the spatial area at a specific time; Indicates that in the current time window In state Instant rewards based on user satisfaction with activities in the area; is a discount factor that controls the importance of future rewards and has a value range of [0, 1]; Indicates that in the current time window From the status Transfer to state probability; Indicates that the next time window is in state the value of Indicates the current state; Indicates the next possible state; Indicates the current time window; Indicates the next time window; Indicates the total number of states; Indicates the number of Sum all j values ​​of .

[0042] Optimal strategy generation: Based on the value function, the optimal area utilization strategy is generated for each time window; ; in, Indicates that in the time window In the area The optimal action when ,guiding spatial layout decision; is a set of possible actions (e.g., different zone configuration options), representing alternative design solutions; Indicates that in the time window Within, users from the area Move to area The conditional probability of , which reflects the impact of the action on the state transition; Indicates returning the parameter that makes the following expression reach the maximum value; Indicates possible actions; Indicates the starting area; Indicates the target area; Indicates the current time window; Indicates the next time window; Indicates the total number of states; Indicates the number of All Value summation; Symbols that represent belonging to a set.

[0043] In specific application scenarios, the time-varying Markov decision process can capture the differences in users' spatial usage patterns at different times. For example, in a small 75-square-meter apartment, the system identifies that users mainly move between the bathroom and kitchen from 6:00 to 8:00 in the morning, mainly in the work area from 8:00 to 12:00 in the morning, frequently use the living room and kitchen areas from 18:00 to 21:00 in the evening, and mainly in the bedroom after 23:00 in the evening. Through these time-varying transition probabilities, the system can more accurately predict the intensity of users' demand for different spatial areas at specific times, thereby providing a time dimension basis for spatial layout optimization. This helps to solve the challenge of small apartments needing to assume different functions at different times. For example, the same area may need to be used as a work area during the day and transformed into a leisure area at night.

[0044] Step 2.3: Calculate the spatial functional requirements score; Based on the biological rhythm model and state transition probability, the functional demand scores of each spatial area at different time periods are calculated. It can be seen that the spatial scoring function Defined as: ; in, represents the spatial scoring function at time t, which quantifies the comprehensive demand intensity for space at a specific time point; represents the biorhythm influence factor, reflecting the degree of influence of the biorhythm on activity i at time t; represents the frequency of activity (i.e., the probability of performing activity i at time t), reflecting the likelihood of the activity occurring at that time point; represents the subjective preference weight (obtained through user questionnaires or feedback), reflecting the user's personal preference for activity i; represents the total number of activity types considered, covering all possible activities that the user may perform; An index representing the activity type; Indicates a point in time; Indicates the number of Sum all i values ​​of ; In addition, this scoring function can quantitatively represent the demand intensity for each functional space in different time periods, providing an important basis for subsequent spatial layout optimization.

[0045] Step 3: Based on the time-series spatial scoring function, the cognitive science modeling method is used to analyze the user's cognitive processing in a multi-tasking environment and calculate the cognitive load index under different spatial layout schemes.

[0046] It includes the following sub-steps: Step 3.1: Task analysis and resource requirement modeling; Apply task analysis techniques to decompose common activities in home scenarios (such as cooking, reading, work, entertainment, etc.) into basic task units, and identify the types of cognitive resources required for each task unit (such as visual attention, auditory attention, working memory, long-term memory retrieval, etc.) and their resource requirements. It should be understood that the task resource requirement matrix Defined as: ; in, Represents the task resource requirement matrix, which describes the requirements of each task for different cognitive resources; Indicates a task Cognitive Resources The demand quantifies the intensity of cognitive resources required to complete a specific task; Indicates the index of the task; An index representing cognitive resources; In addition, the matrix is ​​obtained by analyzing the key behavioral characteristics of users when performing various tasks (such as gaze patterns, operation sequence, error rate, etc.).

[0047] Step 3.2: Construction of ACT-R cognitive computing model; Based on the ACT-R (Adaptive Control of Thought-Rational) cognitive architecture, we constructed a model of users' cognitive processing in a multitasking environment. It should be noted that ACT-R is a computational model that simulates human cognitive processes and includes components such as the perception module, the motor module, the goal module, and the declarative module. With particular attention to working memory load and attention allocation mechanisms, we simulated the task execution process using the following cognitive processing equation: ; in, Indicates the task at time t The activation level reflects the salience of the task in the cognitive system; Indicates a task The basic activation strength reflects the intrinsic importance of the task; Indicates environmental cues The salience at time t reflects the intensity of the external environment's prompts for the task; Indicates clues Task The strength of the association reflects the relevance of environmental cues to the task; Indicates other tasks The activation level reflects the cognitive occupation of the competing task; Indicates a task Task The interference intensity quantifies the degree of mutual interference between tasks; Indicates a point in time; Indicates the index of the current task; An index representing environmental cues; Represents all environmental cues Sum, the value range is 1≤ ≤ ,in is the total number of environmental cues; For all other tasks sum; Indicates the index of other tasks, the value range is 1≤ ≤ , ≠ .

[0048] The specific implementation of the ACT-R cognitive computing model includes the following core components and processing steps: Perceptual Module: responsible for processing visual and auditory input information; Visual buffer: stores information about visual objects in the current field of view; Auditory buffer: stores sound information in the current environment; Visual Location Search: ; in, Represents the visual position activation value based on the feature set at coordinate (x, y); Representation characteristics The weight reflects the importance of the feature; Indicates the matching degree between feature i and position (x, y); The index representing the feature; For all features Summation.

[0049] Visual Object Recognition: ; in, Represents the visual recognition activation value of object obj; Representation characteristics The weight of reflects the importance of the feature in the recognition process; Representation characteristics The degree of matching with object obj; The index representing the feature; For all features Summation.

[0050] Declarative Module: stores and retrieves factual knowledge from long-term memory; Memory retrieval equation: ; in, Represents memory items The degree of activation determines the likelihood that the memory will be retrieved; Represents a memory item The basic activation level reflects the usage history; is the context association weight, which reflects the association strength between the current context and the memory; Current situation clues and memory items The matching degree quantifies the cueing effect of the context on memory retrieval; is random noise, simulating the randomness of memory retrieval; An index representing situational cues; Indicates that all situational clues sum; Represents the index of a memory item.

[0051] Basic activation calculation: ; in, Represents memory items The basic activation level reflects the usage history; Indicates the past The time of the first use of this memory is used to calculate time decay; is the time decay parameter, which controls the decay rate of memory over time; represents the natural logarithm function; Indicates the time point index of past use memory; Indicates that all Value summation; express -d power.

[0052] Extraction probability: ; in, Retrieve memory items probability; Represents memory items activation level; Represents memory items activation level; is the temperature parameter, which controls the randomness of the selection process; represents a natural constant, approximately equal to 2.71828; Indicates the index of the current memory item; Represents the index of all memory items; Indicates that all memory items Summation.

[0053] Goal Module: maintains the status of the current goal and sub-goals; Target buffer: stores the currently activated target state; Target stack: manages switching and recovery of multi-level targets; Target Activation Decay: ; in, Indicates that the target is at time The activation intensity reflects the significance of the target in the cognitive system; is the initial activation intensity, which indicates the importance of the goal when it is set; is the decay parameter, which controls the decay rate of the target over time; Indicates the time since the goal was set; Represents a natural constant, approximately equal to 2.71828.

[0054] Production System: Apply condition-action rules to control cognitive processes; Product matching calculation: ; in, Indicates product form The utility of , which determines the possibility of the rule being selected; It represents the expected return of the product, reflecting the expected value of executing the rules; represents the execution cost, reflecting the cognitive resources required to execute the rules; Representation and Target The strength of association, which quantifies the relevance of the rule to the target; Indicates the target Importance, reflecting the priority of the goal in the current cognitive state; Indicates the index of the current product; Indicates the index of the target; For all targets Summation.

[0055] Conflict Resolution: ; in, Indicates product selection probability; Indicates product form The utility of Indicates product form The utility of is the temperature parameter, which controls the randomness of the selection process; represents a natural constant, approximately equal to 2.71828; Indicates the index of the current product; Represents the index of all product formulas; For all products Summation.

[0056] Multi-tasking execution model: Task Characterization: ; in represents the goal set, which includes all the goals related to the task; Express resource requirements, listing the cognitive resources needed to complete the task; Represents the execution product set, which contains all the rules for executing tasks. Indicates the index of the task.

[0057] Task switching cost calculation: ; in, Indicates that from the task Switch to Task Internal cognitive costs; Represents a resource requirement set and The overlap of two sets quantifies the similarity between them. Represents the target set and The overlap is calculated as the size of the intersection of the two sets divided by the size of the union; Represents a product set and The overlap is calculated as the size of the intersection of the two sets divided by the size of the union; The weight coefficient of the target switching cost reflects the importance of the target change; The weight coefficient of resource switching cost reflects the importance of resource reconfiguration; The weight coefficient of the product switching cost reflects the importance of the change in execution rules. The index number of the starting task; Indicates the index number of the target task.

[0058] In specific application scenarios, the ACT-R cognitive computing model can simulate the cognitive load and task execution efficiency of users in indoor multi-tasking environments. For example, in a space where an open kitchen and living room are combined, the system simulates a scenario in which users are cooking and supervising children's activities at the same time. Through cognitive model analysis, the system found that when there is a visual obstruction between the stove and the children's activity area, users need to frequently switch their attention focus, resulting in an increase in working memory load and a decrease in task execution efficiency. On the contrary, when the design ensures a good visual channel between the two functional areas, users can maintain a "divided attention" state and monitor the two areas at the same time, significantly reducing cognitive load. Similarly, for home office areas that need to perform multiple work tasks at the same time, the system simulates resource competition under different layouts and finds that arranging reference materials and digital devices within the user's optimal field of view can reduce the memory retrieval burden and improve work efficiency by about 25%.

[0059] Step 3.3, task switching cost calculation; Based on the simulation results of the ACT-R model, the switching cost function between different tasks is calculated This function indicates that the layout Under the conditions, from the task Switch to Task The cognitive cost generated is expressed mathematically as follows: ; in, Indicates that the layout Under the conditions, from the task Switch to Task the cognitive costs incurred; Indicates a task and Physical distance of the location; Indicates a task and The degree of heterogeneity in cognitive resource demands; Indicates from position arrive Visual conversion costs (such as gaze shift, posture adjustment, etc.); 、 and represent the weights of physical distance, cognitive difference and visual transformation in task switching cost respectively; Indicates a task Location Indicates a task Location Indicates the starting task; Indicates the target task; Represents the spatial layout plan, is the index number of the starting task, The index number of the target task.

[0060] Step 3.4, comprehensive calculation of cognitive load index; Calculate the comprehensive cognitive load index under a specific spatial layout scheme based on task switching cost and task execution probability : ; in, Representation layout Next task set Cognitive load index; Indicates a task and The probability of sequential execution (obtained from the user behavior data in step 1); Indicates that in the layout Conditions from the task Switch to Task cognitive costs; Indicates the total number of tasks; Indicates the tasks; Indicates the tasks; Represents a spatial layout plan; Represents a task set; is the index number of the starting task, is the index number of the target task. This index provides a quantitative assessment of the impact of spatial layout on user cognitive load. A lower index value indicates that the layout is more conducive to reducing user cognitive burden.

[0061] Step 4: Based on the spatial scoring function and cognitive load index, the improved ant colony algorithm is applied to generate the optimal spatial layout solution that meets multiple objectives.

[0062] It includes the following sub-steps: Step 4.1: Parametric representation of spatial layout; The indoor space is divided into grid units and the spatial layout is represented by parametric coding. It should be understood that the layout decoding ,in 、 、 Represents the parameter sets of functional areas 1, 2, and m, including location coordinates, area size, shape characteristics, and orientation. Indicates the total number of functional areas.

[0063] Step 4.2: Multi-objective function construction; A multi-objective optimization function is constructed, comprehensively considering objectives such as spatial utility, cognitive adaptability, aesthetic value, and construction cost. As can be seen, the specific objective function includes: Spatial utility function : Measure layout The mathematical expression for space utilization efficiency is: ; in, Representation layout The spatial utility function value of ; Represents the functional area at time t Needs score (from step 2); Representation layout Functional area at time t Actual usability ratings; Indicates a point in time; The index representing the functional area; Indicates the total number of functional areas; Represents a spatial layout plan.

[0064] Cognitive load function : Evaluate layout The impact on user cognitive load is expressed mathematically as follows: ; in, Representation layout The cognitive load function value of Representation layout The task set for the next period t Cognitive load index (from step 3); represents the weight coefficient of time period t; Indicates a point in time; represents the task set in time period t; Represents the spatial layout scheme. Note that the negative sign is used here because the optimization goal is to minimize cognitive load.

[0065] Aesthetic value function : Evaluate layout The aesthetic value of is expressed mathematically as follows: ; in, Representation layout The aesthetic value function value of ; represents the symmetry score of the space; Indicates the balance score of the space; Indicates the harmony score of the space; 、 and They respectively represent the importance of symmetry, balance and harmony in aesthetic evaluation; Represents a spatial layout plan.

[0066] Construction cost function : Evaluate implementation layout The required renovation cost is expressed mathematically as follows: ; in, Representation layout The construction cost function value; Indicates functional area renovation costs; Indicates area and Cost of renovation of connections between; Indicates the total number of functional areas; Represents the spatial layout plan. Again, the negative sign is taken here because the optimization goal is to minimize the cost, The index representing the functional area; Indicates the The index of a functional area.

[0067] Step 4.3: Improve the ant colony algorithm to achieve multi-objective optimization; The improved ant colony algorithm (ACO) is used to solve the above multi-objective optimization problem. This improved algorithm is based on the classic ant colony algorithm and introduces the Pareto optimization principle and adaptive pheromone update strategy, which can effectively balance multiple optimization objectives.

[0068] The specific implementation of the improved ant colony algorithm includes the following steps: The problem states: The spatial layout problem is modeled as a graph theory problem, where nodes represent spatial grids and ant paths represent possible spatial layout solutions. Defining multiple pheromone matrices , corresponding to multiple optimization objectives .

[0069] initialization: Randomly generate an initial layout solution set ,in 、 、 Respectively represent the 1st, 2nd, and nth initial layout schemes, Indicates the number of initial layout solutions Initialize pheromone concentration ,in Indicates the initial moment from position To location On target The pheromone concentration on represents the initial pheromone concentration constant, Indicates the starting position, Indicates the target location, The index representing the optimization target; Initialize the non-dominated solution set ,in Represents a set of Pareto non-dominated solutions; Ant mobile and solution build: For each ant m, a layout solution is generated based on multi-target pheromone and heuristic information. ; During the layout construction process, the probability of an ant choosing the next position is: ; in, It's an ant From location Move to position probability; is the target corresponding pheromone concentration; is heuristic information (e.g., spatial fitness); is the target The weight of is the weight of the heuristic information; It's an ant In position The feasible neighborhood of is the total number of optimization objectives; Indicates location A location in the feasible neighborhood of ; Indicates the current location; Indicates the target location; Indicates the index of the ant; The index representing the optimization target.

[0070] Layout adjustments and fixes: Adjust the original layout generated by the ants to ensure that the spatial constraints are met; Apply heuristic rules to fix unreasonable layouts (such as eliminating functional area fragmentation, adjusting channel width, etc.).

[0071] Multi-objective evaluation: For each layout solution Calculate multiple objective function values ; Identify non-dominated solutions based on Pareto dominance relations and update the set of non-dominated solutions .

[0072] Adaptive Pheromone Updates: For each target k, according to the objective function value Update pheromones ; in, represents the pheromone concentration corresponding to target k in the next iteration; Indicates the target of the current iteration corresponding pheromone concentration; is the pheromone evaporation rate; is on target A collection of well-behaved ants; It's an ant On target Pheromone increment on; Indicates the current iteration number; Indicates the starting position; Indicates the target location; The index representing the optimization target; Indicates the index of the ant; Indicates that all Well-behaved ants Summation.

[0073] Pheromone increment calculation: ; in, It's an ant On target Pheromone increment on; is with the goal Related constants; It's an ant Generated layout solution On target Function value on ; Indicates from position To location Path; Represents ants Generated layout solution; Indicates the starting position; Indicates the target location; The index representing the optimization target; Indicates the index of the ant.

[0074] Dynamic weight adjustment: Dynamically adjust the target weights based on the performance distribution of the ant population on each target : ; in, Indicates the target in the next iteration The weight of Indicates the target of the current iteration The weight of is the target the coefficient of variation within the current population; is the average of all target coefficients of variation; is the adjustment factor; Indicates the current iteration number; The index representing the optimization target.

[0075] Pareto front construction: Sort all solutions using the fast non-dominated sorting algorithm; Update Pareto front solution set ; A crowding distance mechanism is applied to maintain the diversity of solutions.

[0076] Termination condition check: Check whether the maximum number of iterations or convergence conditions have been reached; If the termination condition is not met, return to step 3 and continue iterating.

[0077] In specific application scenarios, the improved ant colony algorithm can effectively solve multi-objective optimization problems for interior space layout. For example, in optimizing the layout of a 75-square-meter two-bedroom apartment, the system simultaneously considered the following objectives: 1) maximizing spatial utility within a 24-hour period; 2) minimizing cognitive load in multitasking scenarios; 3) ensuring aesthetic harmony of the decorating styles; and 4) keeping the renovation budget within 120,000 yuan. After 2,000 iterations, the algorithm generated a Pareto front containing 15 non-dominated solutions, representing the optimal layout options under different objective weights. One typical solution achieved a 28% increase in spatial utility, a 32% reduction in cognitive load, and a good aesthetic score of 78 / 100, while keeping the renovation cost below 108,000 yuan.

[0078] Specifically, the solution achieves multi-objective balance through the following key optimizations: The living room and study are designed as semi-open spaces, and movable partitions are used to achieve functional conversion, so as to adapt to work and entertainment needs at different times. Optimize the flow of traffic between the kitchen and the dining room to reduce cross-interference and lower the cognitive load of cooking and dining activities The bedroom layout takes biorhythm factors into consideration, with the head of the bed facing away from noise sources and windows positioned to optimize the entry of natural light and improve sleep quality. The centralized storage space reduces visual clutter while increasing the available activity area.

[0079] Step 4.4: Interactive optimization of layout scheme; Develop an interactive optimization interface that allows users to browse and interactively adjust the set of non-dominated solutions recommended by the system. Users can adjust the weight preferences of various optimization objectives, and the system will then update and display the optimal solution that meets the new preferences. For example, when the user increases the weight of "spatial utility", the system will automatically recommend a layout solution with higher space utilization; when the user increases the weight of "cognitive load", the system will recommend a solution that is more conducive to reducing cognitive burden. During this process, the system will also visualize the quantitative performance of different layout solutions on various objective functions to help users understand the impact of design decisions on multidimensional objectives.

[0080] Step 5: Based on user feedback on the optimal space layout plan and continuously monitored behavioral data, optimize the space layout model and generate a detailed decoration implementation plan.

[0081] It includes the following sub-steps: Step 5.1: User feedback integration and model tuning; Build a user feedback collection and processing mechanism to integrate users' subjective evaluation and preference adjustments of layout solutions into the optimization model. It should be understood that feedback data includes: ① User satisfaction rating of candidate solutions; ② Adjustment requirements for specific functional areas; ③ The degree of preference for a balance between aesthetics and practicality.

[0082] In addition, the system uses the Bayesian update method to convert user feedback into adjustments to model parameters: ; in, and Represent the model parameters before and after the update respectively; represents the learning rate; Represents model output and user feedback The gradient of the loss function between represents the loss function; Represents the model output; Indicates user feedback; In addition, through repeated iterations, the model gradually adapts to the user's personalized needs, improving the customization level of the solution.

[0083] Step 5.2: Dynamic spatial adaptability verification; To verify the dynamic adaptability of the optimized layout in daily use, the system constructs a time-series virtual usage scenario simulator. Based on the user behavior pattern library in step 1, this simulator simulates the complete process of users using the space within a 24-hour period, records potential conflicts, inconveniences, and lack of adaptability, and calculates the space satisfaction score for each period. It should be noted that the adaptability verification formula is: ; in, Representation layout Dynamic adaptability score; represents the spatial demand intensity at time t; Representation layout Carrying capacity at time t; Indicates the total duration of the simulation; Indicates a point in time; represents the smaller value of space demand and carrying capacity at time t; Represents the spatial layout scheme. This indicator measures the degree to which the layout scheme satisfies time-varying requirements, with higher scores indicating greater adaptability.

[0084] Step 5.3: Iterate spatial layout optimization; Based on user feedback and dynamic adaptability verification results, the system iteratively optimizes the layout solution. This optimization process employs an incremental adjustment strategy, retaining user-approved layout elements while focusing on local optimization in problematic areas to avoid drastic changes to the global solution. Furthermore, this iterative optimization employs a reinforcement learning framework, treating layout adjustments as a continuous decision-making process. This approach achieves incremental improvement by maximizing long-term rewards (user satisfaction and adaptability score).

[0085] Step 5.4, generate detailed decoration implementation plan; For the finalized space layout plan, the system generates detailed decoration implementation guidance documents, including: ① Functional area division diagram and area allocation table; ② Furniture size, location and orientation specifications; ③Recommendations on decoration material selection (based on functional requirements and aesthetic coordination); ④ Layout plan of lighting, power supply, network and other infrastructure; ⑤ Requirements for ensuring the circulation path and key dimensions of the traffic flow.

[0086] It should be noted that the implementation plan also includes three-dimensional visual rendering effects to help users and decoration construction personnel intuitively understand the configuration intentions and ensure accurate implementation during the implementation process.

[0087] Real-world application examples of this implementation: This implementation method has been applied in a small two-bedroom apartment of 75 square meters in Beijing. The specific application process and results are as follows: Application scenario description: This 75-square-meter residence, shared by a young couple and a three-year-old child, originally consisted of six functional areas: an entrance hall, living room, master bedroom, second bedroom (children's room), kitchen, and bathroom. Before implementing this technology, users faced the following pain points: ① The couple needed to work remotely from home, but lacked a dedicated workspace; ② The kitchen was cramped, making it difficult to simultaneously cook and watch the child; ③ Insufficient storage space; and ④ Conflicting functional requirements for the space at different times of day, such as a conflict between a daytime work area and an evening leisure area.

[0088] Implementation process example: Data collection and user behavior modeling: Using smart wristbands and indoor sensor networks, we collected data from users' daily lives for 30 days, averaging 15,840 data points per day. Through data preprocessing and feature extraction, we identified typical behavioral patterns of users' families on weekdays and weekends: Weekdays (Monday to Friday): From 6:30 to 8:30 in the morning, the main activity areas are the bathroom, kitchen and living room, and the types of activities include washing, making breakfast, eating and preparing for kindergarten; from 8:30 to 17:30, parents work remotely in the living room, and 60% of the time requires video conferencing or online collaboration; from 17:30 to 20:00, the main activity areas are the kitchen and living room, and the types of activities include cooking, eating and family interaction; from 20:00 to 23:00 is family leisure time, mainly in the living room; after 23:00, activities are mainly in the bedroom.

[0089] Rest days (Saturday and Sunday): Activity time is more flexible, but 9:00-11:00 in the morning and 15:00-18:00 in the afternoon are the peak times for family members to engage in joint activities, mainly in the living room and kitchen; lunch break is approximately 13:00-14:30, and activities are mainly carried out in the bedroom.

[0090] By analyzing 30 days of user location sequence data, the improved LSTM algorithm successfully identified seven typical activity patterns and 31 sub-activity types, and calculated the spatiotemporal distribution characteristics of each activity pattern. For example, the system identified that 7:00-7:30 a.m. on weekdays is the peak time for children and parents to use the restroom simultaneously, and the frequency of spatial conflict during this period is 76%.

[0091] Mapping biological rhythms and spatial requirements: Based on user behavior data, a time-varying Markov decision process (TV-MDP) model was constructed to dynamically change spatial functional demand over a 24-hour period. The demand scores for the main functional areas (living room, master bedroom, second bedroom, kitchen, and bathroom) in 24 time windows (one window per hour) showed the following: Living room area: The highest score for work function demand is from 9:00-17:00 on weekdays (average 85 points / 100 points), and the highest score for dining and family interaction function demand is from 18:00-21:00 (average 88 points / 100 points).

[0092] Kitchen area: Peak usage hours are 7:00-8:00 and 18:00-19:00 on weekdays (average demand score 92 / 100 points), and visual connectivity with the living room is required.

[0093] Master bedroom area: It needs to support both sleeping and temporary work functions on weekdays, and focus more on leisure functions on weekends.

[0094] Biorhythm analysis shows that the user's family members have the highest cognitive acuity between 8:00-10:00 and 19:00-21:00 in the morning, which is suitable for work or study activities that require high concentration; while the level of physical fatigue is the highest between 14:00-15:00, which is suitable for rest.

[0095] Cognitive load assessment and task flow analysis: Through the ACT-R cognitive computing model, the user's cognitive processing in a multi-tasking environment is quantitatively analyzed. Key task scenarios include: Remote work and childcare scenarios: When parents need to simultaneously look after their children playing in the living room while working in the living room, the cognitive load index under the traditional spatial layout is 78 / 100 (high), mainly due to frequent switching of visual attention and auditory interference.

[0096] Cooking and socializing scenario: When a family member prepares dinner in the kitchen and needs to communicate with family members in the living room at the same time, the cognitive load index of the traditional closed kitchen layout is 82 / 100 (high), mainly due to the distraction and reduced communication efficiency caused by visual partitions.

[0097] Calculations of task switching costs showed that, under the original spatial layout, the switching cost between work and childcare was 0.68 (on a scale of 0-1, with higher switching costs indicating a greater switching burden), and the switching cost between cooking and socializing was 0.72. These high switching costs were primarily due to physical isolation and lack of visual access.

[0098] Multi-objective optimization spatial layout generation: Applying the improved ant colony algorithm and Pareto optimization principle, integrating the analysis results of the previous steps, the following multi-objective optimization function is set: Spatial utility function: maximize the utilization efficiency of each functional area within a 24-hour period; Cognitive load function: minimizes cognitive burden in multitasking scenarios; Aesthetic value function: ensure that the aesthetics of the space is not lower than the original level; Construction cost function: Control the decoration budget within 100,000 yuan.

[0099] After 3,000 iterative calculations, the system generated a Pareto front of 12 non-dominated solutions. The final recommended optimization solution included the following key design points: The living room was re-planned as a multifunctional space of "daytime work area + evening leisure area", and flexible space conversion was achieved through lightweight movable partitions and multifunctional furniture.

[0100] The kitchen is changed to a semi-open design, and a bar-style partition is added between it and the living room, which not only maintains visual connectivity but also effectively controls the spread of kitchen fumes.

[0101] The master bedroom is designed as a composite space of "sleeping area + small work area". The work area uses a foldable desktop that can be completely stored when not in use.

[0102] Customized storage solutions throughout the house utilize wall space and built-in storage in furniture to increase storage capacity by 35%.

[0103] Based on the characteristics of biological rhythms, natural light and artificial lighting systems are optimized to provide suitable lighting environments at different times.

[0104] In terms of weight distribution, the multi-objective optimization algorithm sets the spatial utility weight to 0.4, the cognitive load weight to 0.3, the aesthetic value weight to 0.2, and the construction cost weight to 0.1 based on user preferences. The final solution generated achieves balanced performance in all objective dimensions.

[0105] Technical effect verification: After the renovation was completed, the technical effectiveness of this implementation was verified through three months of continuous tracking monitoring and user feedback collection, focusing on two key indicators: space utilization efficiency and cognitive load. The results are shown in Tables 1 and 2: Table 1: Data table of space utilization efficiency improvement effect

[0106] Table 2: Cognitive load reduction effect data table

[0107] In addition to the two core technical effects mentioned above, users also reported the following significant improvements: Family members' interaction time has significantly increased, and parent-child relationship satisfaction, work efficiency, and remote work satisfaction have all improved; If there are no functional adaptability issues after one year of use, no secondary renovation is required; With ample storage space at home, you spend less time searching for items.

[0108] The application of this implementation method verifies the feasibility and effectiveness of the large-model-based intelligent design method for interior decoration in actual scenarios, especially its significant technical effects in terms of space optimization, cognitive load reduction, all-weather functional adaptability and personalized customization, providing an effective solution for the intelligent design of multi-functional spaces in small apartments.

[0109] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of equivalent embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A large-scale model-based intelligent design method for interior decoration, characterized in that: The following steps are involved: Collect user activity data, build user behavior models, and generate a personalized activity pattern library; Based on the personalized activity pattern library, a time-varying Markov decision process is applied to construct spatial functional demand models for different time periods and generate a time-series spatial scoring function. Based on the temporal spatial scoring function, cognitive science modeling methods are used to analyze the user's cognitive processing in a multi-tasking environment and calculate the cognitive load index under different spatial layout schemes. According to the spatial scoring function and cognitive load index, an improved ant colony algorithm is applied to generate the optimal spatial layout solution that meets multiple objectives. Based on user feedback on the optimal space layout plan and continuously monitored behavioral data, the space layout model is optimized and a detailed decoration implementation plan is generated.

2. The method for intelligent interior decoration design based on a large model according to claim 1, characterized in that: The collection of user activity data includes: collecting the user's location movement trajectory, activity status and physiological parameters through smart wearable devices, and monitoring the user's specific location, stay duration and activity type indoors through indoor sensor networks.

3. The method for intelligent design of interior decoration based on a large model according to claim 1, characterized in that: The steps of constructing the user behavior model include: Using the improved long short-term memory network algorithm, the collected raw data are subjected to feature extraction and pattern recognition; Divide user activities into multiple categories and record the spatiotemporal distribution characteristics, frequency characteristics, and correlation characteristics of each category of activities; Extract activity patterns within a 24-hour period, including peak and trough times for different activities.

4. The method for intelligent interior decoration design based on a large model according to claim 1, characterized in that: The steps of constructing spatial functional demand models for different time periods using the time-varying Markov decision process include: Construct an individual biorhythm model and define biorhythm influencing factors; Divide the indoor space into multiple functional areas and build a probability model of users' spatial state transitions at different times; Based on the biological rhythm model and state transition probability, the functional demand scores of each spatial area at different time periods are calculated.

5. The method for intelligent design of interior decoration based on a large model according to claim 1, characterized in that: The steps of the cognitive science modeling method for analyzing the user's cognitive processing in a multi-tasking environment include: Apply task analysis techniques to decompose common activities in home settings into basic task units and identify the types and amounts of cognitive resources required for each task unit; Based on the ACT-R cognitive architecture, a user's cognitive processing model in a multi-tasking environment is constructed; Calculate the switching cost function between different tasks; Based on the task switching cost and task execution probability, the comprehensive cognitive load index under a specific spatial layout scheme is calculated.

6. The method for intelligent design of interior decoration based on a large model according to claim 1, characterized in that: The steps of applying the improved ant colony algorithm to generate an optimal spatial layout solution that meets multiple objectives include: The indoor space is divided into grid units and the spatial layout is represented by parametric coding. Construct a multi-objective optimization function that comprehensively considers spatial utility, cognitive adaptability, aesthetic value, and construction cost objectives; An improved ant colony algorithm that introduces the Pareto optimization principle and adaptive pheromone update strategy is used to solve multi-objective optimization problems. Develop an interactive optimization interface that enables users to browse and interactively adjust the set of non-dominated solutions recommended by the system.

7. The method for intelligent design of interior decoration based on a large model according to claim 6, characterized in that: The multi-objective optimization function includes: Spatial utility function, which measures the efficiency of the layout in space utilization; Cognitive load function, which evaluates the impact of layout on user cognitive load; Aesthetic value function, which evaluates the aesthetic value of the layout; A construction cost function that evaluates the renovation costs required to realize the layout.

8. The method for intelligent design of interior decoration based on a large model according to claim 1, characterized in that: The steps to optimize the space layout model and generate a detailed renovation implementation plan include: Build a user feedback collection and processing mechanism to integrate users' subjective evaluation and preferences of layout solutions into the optimization model; Build a time-series virtual usage scenario simulator to verify the dynamic adaptability of the optimized layout in daily use; Adopting incremental adjustment strategies and reinforcement learning frameworks to iteratively optimize layout solutions; Generate detailed decoration implementation guidance documents and 3D visual rendering effects.

9. The method for intelligent interior decoration design based on a large model according to claim 1, characterized in that: The improved ant colony algorithm comprises the following steps: Calculating multiple objective function values ​​for each layout solution; Identify non-dominated solutions based on the Pareto dominance relationship and update the set of non-dominated solutions; Update pheromone for each target according to the objective function value; Dynamically adjust the target weights based on the performance distribution of the ant population on each target; All solutions are sorted using the fast non-dominated sorting algorithm and the Pareto front solution set is updated.

10. An intelligent interior decoration design system based on a large model, characterized in that: The method for executing the large model-based intelligent design method for interior decoration according to any one of claims 1 to 9 comprises: Data collection module, used to collect user activity data, build user behavior models, and generate personalized activity pattern libraries; The demand mapping module is used to construct spatial functional demand models for different time periods based on user behavior data and apply the time-varying Markov decision process to generate a time-series spatial scoring function; The cognitive assessment module is used to analyze the user's cognitive processing in a multi-tasking environment through cognitive science modeling methods and calculate the cognitive load index under different spatial layout schemes; An optimization generation module is used to apply an improved ant colony algorithm to generate an optimal spatial layout solution that meets multiple objectives based on the results of biorhythm space demand and cognitive load assessment; Implementation guidance module, used to optimize space layout model and generate detailed decoration implementation plan based on user feedback and continuously monitored behavioral data.

Images