Auxiliary hospital space design method and device based on building energy consumption model

Abstract

The present invention is an auxiliary hospital space design method and device based on a building energy consumption model, including: obtaining a natural language description of the initial hospital space layout diagram and functional requirements; extracting features from the space layout diagram and performing semantic analysis on the functional requirements description; inputting the spatial structure feature matrix into the energy consumption analysis model, combining the demand semantic vector, predicting the energy consumption performance of each functional area, and generating an energy consumption assessment diagram; obtaining an energy-optimized space design scheme based on the energy consumption assessment diagram and functional requirements; and simulating and verifying the energy-optimized space design scheme. The present invention realizes the intelligentization, optimization, and dynamic adjustment of hospital space design by integrating advanced technologies such as multimodal deep learning, generative adversarial networks, building energy consumption simulation, adaptive optimization algorithms, and multi-objective decision support systems.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving buildings, and in particular to an auxiliary hospital space design method and device based on a building energy consumption model. Background Art

[0002] In recent years, the application of building energy consumption models in hospital space design has gradually increased, aiming to optimize building design by simulating and predicting energy consumption. However, the existing technology has some limitations.

[0003] Traditional energy consumption models are usually based on static parameters and cannot accurately reflect the complex and dynamic environment of hospitals, where personnel flow is frequent and equipment utilization is high. These models often only consider a single energy consumption factor, such as lighting or air conditioning, and lack a comprehensive analysis of multiple energy consumption factors. In spatial layout design, existing technologies do not fully understand and integrate user needs and functional requirements, and cannot fully meet the diverse functional needs of hospitals, such as close connections between departments, efficient personnel flow, and patient comfort.

[0004] These problems make it difficult to strike a balance between energy consumption optimization and functional satisfaction in the design solution, limiting the overall performance improvement of hospital buildings. It is urgent to provide a new solution to improve design efficiency, optimize energy use, meet complex functional requirements, and overcome the shortcomings of existing technologies. Summary of the invention

[0005] In order to overcome the shortcomings of the existing technology, the present invention proposes an auxiliary hospital space design method and device based on the building energy consumption model. By integrating advanced artificial intelligence and architectural technology, it realizes all-round optimization of hospital space design and significantly improves design efficiency, energy efficiency performance and user experience.

[0006] To achieve the above object, the present invention provides an auxiliary hospital space design method based on a building energy consumption model, comprising the following steps:

[0007] Step 1: Obtain the natural language description of the initial hospital space layout and functional requirements provided by the user;

[0008] Step 2: Extract features from the spatial layout diagram to obtain the spatial structure feature matrix, and perform semantic analysis on the functional requirement description to obtain the requirement semantic vector;

[0009] Step 3: Input the spatial structure feature matrix into the energy consumption analysis model, combine it with the demand semantic vector, predict the energy consumption performance of each functional area, and generate an energy consumption assessment diagram;

[0010] Step 4: According to the energy consumption assessment diagram and functional requirements, use the optimization algorithm to adjust the space layout and obtain the energy-optimized space design plan;

[0011] Step 5: Simulate and verify the energy-optimized space design to ensure that it achieves optimal energy performance while meeting functional requirements.

[0012] Furthermore, step 1 is as follows:

[0013] Step 1.1: Obtain the initial spatial layout of the hospital, digitize the layout, and convert it into a computer-processable format, such as a vector diagram or a three-dimensional model;

[0014] Step 1.2: The user describes the functional requirements of the hospital in detail in natural language, including department settings, personnel flow, special medical equipment requirements, future expansion plans, etc. The natural language description is preliminarily sorted to ensure that the information is complete and accurate;

[0015] Step 1.3: Normalize the spatial layout diagram, unify the scale and coordinate system, and extract the geometric features and topological structure of the building;

[0016] Step 1.4: Perform text preprocessing on the natural language description, including word segmentation, part-of-speech tagging, and named entity recognition to extract key functional requirements and constraints.

[0017] Furthermore, step 2 is as follows:

[0018] Step 2.1: Based on the traditional convolutional neural network (CNN), spatial position encoding is added to integrate the relative position and direction information of spatial elements into the convolution kernel to enhance the model's perception of spatial structure;

[0019] Step 2.2: Combine CNN with graph convolutional network, use GCN to process the topological structure of spatial layout graph and capture the connection relationship between spatial units;

[0020] Specifically include:

[0021] In the convolution operation, the convolution kernel not only considers the value of the neighborhood pixels, but also combines their spatial positions to form position-sensitive feature extraction.

[0022] The spatial layout graph is represented as a graph structure, where nodes represent spatial units and edges represent connection relationships, and GCN is used to extract topological features.

[0023] Step 2.3: Input the preprocessed spatial layout map into SSP-CNN to extract multi-level spatial feature maps, including information such as spatial morphology, connection relationship, and functional partitioning.

[0024] Step 2.4: Based on the pre-trained BERT model, fine-tune it using a large amount of professional corpus in the field of medical architecture to enable the model to better understand professional terminology and specific requirements.

[0025] Step 2.5: In semantic encoding, increase attention to key demand words and use the attention mechanism to highlight important information.

[0026] Specifically include:

[0027] A large amount of text is collected from medical building design documents, specifications and professional papers to build a domain corpus.

[0028] Use the domain corpus to continue training the BERT model and adjust the model parameters to adapt it to the language characteristics of a specific domain.

[0029] During the encoding process, higher weights are given to the identified key demand words to enhance their influence in the semantic vector.

[0030] Step 2.6: Input the preprocessed natural language requirements into Domain-Adaptive BERT to obtain a high-dimensional requirement semantic encoding feature vector, which contains functional requirements, constraints, and priority information.

[0031] Step 2.7: When fusing spatial features and semantic features, a bidirectional attention mechanism is established so that the information of the two modalities can influence and complement each other.

[0032] Step 2.8: Fusion is performed at different feature levels, taking into account high-level abstract features while retaining low-level detail information.

[0033] Specifically include:

[0034] The attention weights of spatial features on semantic features and vice versa are calculated to obtain weighted feature representations.

[0035] The weighted spatial and semantic features are concatenated or weighted summed to form a fused feature vector.

[0036] Repeat the above fusion process at different levels of the model (such as convolutional layer, fully connected layer) to ensure comprehensive fusion of information.

[0037] Step 2.9: Input the spatial feature map and the requirement semantic encoding feature vector into MIAF-Net to generate a fused feature representation and establish a deep association between spatial layout and functional requirements.

[0038] Furthermore, step 3 is as follows:

[0039] Step 3.1: Introduce conditional information (fused feature representation) in the generator and discriminator so that the generation process is guided by specific requirements.

[0040] Step 3.2: The generator outputs multi-scale layout solutions, and the discriminator evaluates the rationality of the layout at different scales; specifically, it includes:

[0041] The generator (G) inputs the fused feature representation and generates a spatial layout diagram that meets the requirements through a series of deconvolution and activation functions.

[0042] The discriminator (D) receives the real layout and the generated layout, as well as the corresponding condition information, and judges the authenticity of the layout and its matching degree with the requirements.

[0043] The loss function design adds conditional matching loss on the basis of traditional GAN ​​loss to ensure that the generated layout meets the functional requirements.

[0044] Step 3.3: Prepare training data, including a large number of excellent hospital design plans and their corresponding functional requirements descriptions, and train the conditional generative adversarial network;

[0045] Step 3.4: Use the real layout and condition information to train the discriminator so that it can distinguish between the real and generated layouts and determine whether the layout meets the conditions;

[0046] Step 3.5: Train the generator to generate layout solutions that are difficult for the discriminator to distinguish and meet the requirements of conditional information

[0047] Step 3.6: Re-input the generated layout solution into the feature extractor and fusion network to obtain a new fusion feature representation;

[0048] Step 3.7: Calculate the difference between the initial fused feature representation and the regenerated fused feature representation as the cycle consistency loss;

[0049] Step 3.8: Input the user's fused feature representation into the trained generator to generate a preliminary spatial layout plan;

[0050] Step 3.9: Use multi-objective optimization strategy to optimize the solution;

[0051] Specifically include:

[0052] Taking into account multiple objectives such as minimization of energy consumption, satisfaction of functional requirements, and space utilization, a multi-objective optimization function is constructed.

[0053] The initially generated layout scheme is optimized using an evolutionary algorithm to obtain a set of Pareto optimal solutions.

[0054] Step 3.10: Output the optimized spatial layout plan, including detailed functional area division and energy consumption assessment data.

[0055] Furthermore, step 4 is as follows:

[0056] Step 4.1: Define multiple evaluation indicators, including spatial functional adaptability, energy efficiency performance, rationality of personnel flow lines, patient comfort, etc. A weighted scoring mechanism is used for each indicator.

[0057] Step 4.1: Adjust the weights of each evaluation criterion through user feedback or simulation calculations to make the evaluation process adaptive to different user needs; specifically include

[0058] Evaluate whether the layout of each functional area meets the specifications and requirements of medical design.

[0059] Based on the generated spatial layout plan, the building energy efficiency is analyzed, the energy consumption of lighting, air conditioning, daylighting, etc. is calculated, and optimization suggestions are put forward.

[0060] Simulate the flow of medical staff, patients, and supplies, evaluate the impact of the layout on the flow of personnel, and ensure reasonable flow lines.

[0061] According to hospital design standards, evaluate the comfort of patients' waiting, treatment and resting spaces, especially in terms of quietness and privacy.

[0062] Introduce machine learning models or expert systems, adjust the weights of each evaluation indicator through feedback data, and optimize the design plan in real time.

[0063] Step 4.2: Use CEM to comprehensively evaluate multiple candidate solutions and assign scores to each solution based on the evaluation criteria;

[0064] Step 4.3: Compare the assessment results with user needs and identify the spatial layout solution that meets the best standards;

[0065] Step 4.4: Use a multi-objective optimization algorithm (such as NSGA-II) to obtain multiple Pareto optimal solutions, and select the solution that best suits user needs through a decision support system (MODSS);

[0066] Step 4.2: Adjust the decision criteria based on user feedback (such as budget, function priority, energy consumption tolerance, etc.) and automatically recommend the optimal solution.

[0067] Specifically, based on the evaluation criteria, one or more spatial layout solutions are selected as the Pareto optimal solution to ensure that the design solution achieves a balance between multiple objectives.

[0068] MODSS decision support combines the personalized needs of users and guides users to choose the most appropriate space layout plan through the decision support system. For example, users can set higher energy efficiency requirements or more compact space layout as the basis for decision making.

[0069] After the user selects a plan, the system provides relevant feasibility reports and optimization suggestions to further guide the decision-making process.

[0070] Furthermore, step 5 is as follows:

[0071] Step 5.1: Use a generative adversarial network (GAN) combined with AutoCAD automation tools to convert the final selected spatial layout plan into detailed architectural design drawings;

[0072] Step 5.2: The generated design drawings include multi-level design documents such as the building general plan, layout drawings of each functional area, equipment configuration drawings, lighting and ventilation drawings, etc.;

[0073] Specifically include:

[0074] The generated space layout plan can be directly combined with AutoCAD to automatically output floor plans, elevations, sections, etc. that meet architectural design specifications.

[0075] GAN generates architectural design drawings of different styles based on the user's layout requirements and makes fine adjustments through AutoCAD.

[0076] Through deep learning models, design drawings of different dimensions are automatically generated, covering multiple aspects such as architecture, structure, equipment, streamlines, etc.

[0077] Step 5.3: Use the deep learning model to simulate the construction process of the hospital building and simulate the construction difficulty, time schedule and cost.

[0078] Step 5.4: Optimize the construction process through simulation results, provide the best construction plan, and reduce uncertainty and risks in construction;

[0079] Specifically include:

[0080] Use construction project management tools such as BIM models combined with deep learning simulation to generate detailed construction process and risk assessment reports.

[0081] Based on the simulation results, reinforcement learning algorithms are used to optimize resource scheduling to ensure the optimization of project progress and cost.

[0082] The use of deep learning models for construction simulation and intelligent optimization has greatly improved the predictability and controllability of the construction process and reduced construction risks.

[0083] Step 5.5: During the construction implementation process, collect feedback information from actual use, automatically adjust the design plan, and perform iterative optimization;

[0084] Step 5.1: Based on user feedback (such as space utilization, energy consumption, comfort, etc.), automatically adjust the layout plan to achieve dynamic optimization.

[0085] Specifically include:

[0086] Data on building use, such as temperature, humidity, and personnel flow, are collected through sensors, smart monitoring equipment, and other equipment.

[0087] Use machine learning algorithms to analyze feedback data, intelligently adjust design plans, and generate updated optimized drawings.

[0088] The device for assisting the hospital space design method based on the building energy consumption model is suitable for any of the above-mentioned assisting hospital space design methods based on the building energy consumption model, including an input module, a cross-modal feature extraction module, a space saliency and demand guidance module, an energy efficiency analysis and optimization module, a decision support and layout generation module, and a construction simulation and feedback adjustment module.

[0089] Furthermore, the input module is used to obtain hospital design requirements (such as functional area requirements, space usage preferences, budget, etc.) input by the user through natural language processing technology; the user inputs the hospital space boundary map through the graphical interface or uploads the existing space layout using a scanned image;

[0090] The cross-modal feature extraction module uses a convolutional neural network to perform deep learning feature extraction on the input apartment boundary map to generate a shallow feature map of the spatial layout; it performs semantic encoding on the user's natural language description and converts the demand into a high-dimensional semantic feature vector through the BERT or Transformer model;

[0091] The spatial saliency and demand guidance module uses multi-scale convolutional pooling and attention mechanisms to focus on areas with spatial saliency, improving the energy efficiency and functional adaptability of spatial layout. It interactively encodes the semantic features of user needs and spatial saliency features to generate demand-guided spatial layout solutions.

[0092] The energy efficiency analysis and optimization module is based on the building energy efficiency model (BEM). It evaluates the energy efficiency of each design scheme, calculates various energy consumption, and optimizes the scheme according to the energy efficiency simulation results and user needs to ensure the best energy efficiency.

[0093] The decision support and layout generation module conducts a comprehensive evaluation based on multi-dimensional evaluation criteria and selects the design solution that best meets the needs; through the combination of GAN and AutoCAD technology, it automatically generates architectural design drawings that meet the specifications, including floor plans, elevations, functional area distribution maps, etc.

[0094] The construction simulation and feedback adjustment module uses deep learning and building information modeling to simulate the construction process and simulate resource scheduling, time management and risk assessment during the construction process. After the design plan is implemented, the actual use of the building is monitored in real time through the Internet of Things technology, feedback data is collected and the design plan is adjusted to form a closed-loop optimization.

[0095] Compared with the prior art, the present invention has the following beneficial effects:

[0096] 1. The present invention provides an auxiliary hospital space design method and device based on a building energy consumption model. By using cross-modal feature extraction and fusion technology, the user's apartment boundary map and natural language design requirements are deeply analyzed and combined to quickly generate an initial design plan; the automated design process reduces manual participation, improves design efficiency, and avoids human errors.

[0097] 2. The present invention provides an auxiliary hospital space design method and device based on a building energy consumption model. Through building energy consumption simulation (BEM) and adaptive optimization algorithm, the energy consumption of different design schemes can be accurately predicted and optimized; the building energy consumption can be optimized in the design stage, thereby reducing operating costs.

[0098] 3. The present invention provides an auxiliary hospital space design method and device based on a building energy consumption model. Through construction simulation and optimization, deep learning and BIM technology are used to simulate the construction process in advance, identify potential risks, optimize the construction plan, reduce uncertainty and resource waste in the construction process, and improve the success rate of the project.

[0099] 4. The present invention provides an auxiliary hospital space design method and device based on a building energy consumption model, which monitors building usage data in real time through the Internet of Things (IoT) technology to form a feedback mechanism; realizes dynamic optimization and adjustment of the design scheme to ensure that the building maintains optimal performance in actual operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0101] Figure 1 It is a schematic diagram of the steps of the present invention

[0102] Figure 2 Schematic diagram of the device of the present invention DETAILED DESCRIPTION

[0103] The technical solution of the present invention will be more clearly and completely explained below through description of preferred embodiments of the present invention in combination with the accompanying drawings.

[0104] Terminology explanation:

[0105] BERT: Transformer model represented by a bidirectional encoder. Transformer is a neural network model based on the attention mechanism, which can effectively handle the long-distance dependency problem of sequence data.

[0106] GAN: Generative Adversarial Network, a deep learning model composed of a generator and a discriminator. The generator is responsible for generating data, and the discriminator is responsible for determining whether the data is true or false. The two compete with each other, and eventually the generator can generate data that is indistinguishable from the real thing.

[0107] MODSS: Multi-Objective Decision Support System

[0108] BIM: Building Information Modeling

[0109] SSP-CNN: Spatial Structure Perceptual Convolutional Neural Network

[0110] MIAF-Net: Multimodal Interactive Attention Fusion Network

[0111] cGAN: Conditional Generative Adversarial Networks

[0112] NSGA-II: Non-dominated Sorting Genetic Algorithm II

[0113] BEM: Building Energy Model

[0114] like Figure 1 As shown, the present invention is specifically:

[0115] Step 1: Obtain the natural language description of the initial spatial layout of the hospital and the functional requirements;

[0116] Step 2: Extract features from the spatial layout diagram to obtain the spatial structure feature matrix, and perform semantic analysis on the functional requirement description to obtain the requirement semantic vector;

[0117] Step 3: Input the spatial structure feature matrix into the energy consumption analysis model, combine it with the demand semantic vector, predict the energy consumption performance of each functional area, and generate an energy consumption assessment diagram;

[0118] Step 4: According to the energy consumption assessment diagram and functional requirements, use the optimization algorithm to adjust the space layout and obtain the energy-optimized space design plan;

[0119] Step 5: Simulate and verify the energy-optimized space design to ensure that it achieves optimal energy performance while meeting functional requirements.

[0120] As a specific implementation method, the user first inputs the floor plan boundary map and the design requirements described in natural language through the interface. The floor plan boundary map can be input through drawing tools, scanning or uploading images, and the design requirements are provided through a text box or voice recognition interface. After receiving these inputs, the system first performs data preprocessing to extract the spatial layout features of the floor plan, such as basic elements such as room boundaries, door and window locations, etc. At the same time, the system performs natural language processing on the design requirements, converts them into structured semantic information, and extracts requirements for spatial functions, usage frequency, and spatial layout relationships. These feature information provides the basis for subsequent feature extraction.

[0121] The processed input data enters the cross-modal feature extraction module. In this module, the system uses convolutional neural networks to perform deep feature extraction on the floor plan to obtain image features of the spatial layout, including room size, positional relationships, and the connection between different spaces. At the same time, for design requirements, the system uses natural language processing technology to process the user's text description and generate semantic vectors to represent the detailed information of the spatial function and design requirements. The system fuses these two different forms of features to produce a joint feature representation, ensuring that the spatial layout and design requirements can be reasonably expressed and optimized during the spatial design process.

[0122] After feature extraction is completed, the system enters the stage of spatial saliency and demand guidance. In this stage, the system uses a multi-scale attention mechanism to saliency the spatial features of the floor plan. By weighting spatial features at different scales, the system can highlight key spatial areas related to design requirements, such as medical function areas, channel areas, etc. Through the cross-modal interactive encoder, the semantic vectors of spatial features and design requirements are further fused to generate a guided feature map. This guided map will serve as the basis for spatial layout optimization in the subsequent spatial optimization process to ensure that the design plan meets the user's requirements.

[0123] The demand-guided spatial feature map will enter the energy efficiency analysis and optimization module. In this module, the system uses the building energy efficiency simulation model to evaluate the energy efficiency of different design schemes, taking into account factors such as the thermal performance of building materials, energy consumption, and indoor air circulation. Through energy efficiency simulation, the system can calculate the energy efficiency performance of each design scheme and then identify potential energy efficiency problems. Based on this evaluation result, the system uses an adaptive optimization algorithm to adjust the design scheme to maximize energy efficiency. This includes optimizing the energy efficiency of the building and reducing energy consumption by adjusting factors such as window size, room layout, and air conditioning system layout.

[0124] After energy efficiency optimization, the system enters the design solution generation stage. The system generates multiple candidate solutions through the decision support module, combining energy efficiency optimization results, spatial functional requirements, comfort and other factors. In this process, the system will generate design drawings through generative adversarial networks. These drawings will meet the functional requirements and energy efficiency goals of the building and ensure feasibility during construction. The generated design drawings are further processed by AutoCAD tools to generate accurate construction drawings for reference by architects and construction personnel.

[0125] The completed design drawings will be simulated through building information modeling (BIM) technology to simulate the entire construction process. BIM technology can help identify potential conflicts between design and construction, plan construction steps in advance, and optimize the construction process. During the actual construction process, the system uses the Internet of Things technology to monitor the data of the construction site in real time, such as equipment usage, material consumption, etc. Through these data feedback, the system can adjust the design and construction plan.

[0126] like Figure 2 As shown in FIG. 1 , the device includes an input module, a cross-modal feature extraction module, a spatial saliency and demand guidance module, an energy efficiency analysis and optimization module, a decision support and layout generation module, and a construction simulation and feedback adjustment module. The device receives the natural language description of the apartment boundary map and the design requirements input by the user. The user inputs the apartment boundary map through the interface, and the system directly obtains the image data by scanning, uploading or drawing tools. For the design requirements, the user inputs the requirements for spatial layout, functional partitioning, frequency of use, comfort, etc. through a text box, voice recognition or natural language processing interface. After preprocessing, the input apartment boundary map and design requirements enter the cross-modal feature extraction module for further analysis. At this stage, the apartment boundary map is image-featured through a model based on a convolutional neural network, and the extracted spatial features include indoor and outdoor boundaries, room partitions, etc. At the same time, the natural language description of the design requirements is semantically encoded through the BERT model, and the requirements in the text are converted into high-dimensional semantic vectors. These feature vectors represent the specific content and spatial requirements of the design requirements, providing guidance for subsequent spatial layout optimization.

[0127] After the cross-modal feature module is extracted, the generated floor plan features and semantic features of the design requirements are transmitted to the spatial saliency and demand guidance module. In this module, the multi-scale pan-spatial attention mechanism is first applied to saliency the extracted floor plan boundary map features. This process highlights the most critical areas in the space for design requirements by weighting the features of each scale. The use of the pan-spatial attention mechanism not only considers the saliency of spatial areas, but also can discover potential design requirements in multi-scale spatial features, which is especially suitable for multi-function and multi-requirement situations in complex hospital building layouts.

[0128] The cross-modal interactive encoder further integrates the spatially significant apartment features with the semantic features of the design requirements, and generates a guiding feature map that comprehensively considers the spatial layout and design requirements through multi-level interactive learning.

[0129] The demand-guided feature map then enters the energy efficiency analysis and optimization module, where the building energy efficiency simulation model performs energy efficiency analysis on the space design. The energy efficiency simulation not only takes into account factors such as the thermal performance of building materials, the lighting of walls and windows, but also the energy consumption of equipment in the building, air circulation and air conditioning systems. Through a comprehensive analysis of the building's energy efficiency, the system can estimate the energy consumption under different layouts and provide data support for the design plan. The adaptive optimization algorithm automatically adjusts the design plan on this basis to generate the most energy-efficient spatial layout.

[0130] The design scheme after energy efficiency optimization will be passed to the decision support and layout generation module. In this module, the system uses a multi-objective decision support system to optimize the architectural design. Through the decision of this module, the system can automatically generate multiple candidate design schemes, and combine the generative adversarial network with the AutoCAD drawing generation module to automatically convert these design schemes into architectural drawings with detailed dimensions and structures. Compared with traditional design methods, GAN can generate highly realistic and innovative design drawings, while AutoCAD ensures the accuracy and practicality of the drawings.

[0131] The construction simulation and feedback adjustment module simulates the construction process through the building information model (BIM) technology. BIM technology can accurately simulate all aspects of building construction, help identify potential construction problems in advance, and optimize the construction process. This module can also monitor the situation on the construction site in real time. The system adjusts the design and construction plan based on the real-time collected data to ensure that the final construction results are consistent with the design goals. The introduction of the feedback mechanism enables the system to dynamically optimize the design plan based on the on-site data during the construction process, adjust the layout in time, and ensure that the implementation effect of the design is consistent with the design goals set initially.

[0132] The above specific implementations are only descriptions of the preferred implementations of the present invention, and do not limit the protection scope of the present invention. Without departing from the design concept and spirit of the present invention, various modifications, substitutions and improvements made by ordinary technicians in this field to the technical solution of the present invention based on the text description and drawings provided by the present invention should all fall within the protection scope of the present invention. The protection scope of the present invention is determined by the claims.

Claims

1. Auxiliary hospital space design method based on building energy consumption model, characterized in that: The following steps are involved: Step 1: Obtain the natural language description of the initial spatial layout of the hospital and the functional requirements; Step 2: Extract features from the spatial layout diagram to obtain the spatial structure feature matrix, and perform semantic analysis on the functional requirement description to obtain the requirement semantic vector; Step 3: Input the spatial structure feature matrix into the energy consumption analysis model, combine it with the demand semantic vector, predict the energy consumption performance of each functional area, and generate an energy consumption assessment diagram; Step 4: According to the energy consumption assessment diagram and functional requirements, use the optimization algorithm to adjust the space layout and obtain the energy-optimized space design plan; Step 5: Simulate and verify the energy-optimized space design to ensure that it achieves optimal energy performance while meeting functional requirements.

2. The auxiliary hospital space design method based on building energy consumption model according to claim 1 is characterized in that: Step 1 is as follows: Step 1.1: Obtain the initial spatial layout of the hospital, digitize the layout, and convert it into a computer-processable format; Step 1.2: The user describes the functional requirements of the hospital in detail in natural language, and the natural language description is preliminarily organized; Step 1.3: Normalize the spatial layout diagram, unify the scale and coordinate system, and extract the geometric features and topological structure of the building; Step 1.4: Perform text preprocessing on the natural language description, including word segmentation, part-of-speech tagging, and named entity recognition to extract key functional requirements and constraints.

3. The auxiliary hospital space design method based on building energy consumption model according to claim 1 is characterized in that: Step 2 is as follows: Step 2.1: Based on the traditional convolutional neural network, i.e. CNN, spatial position encoding is added to incorporate the relative position and direction information of spatial elements into the convolution kernel; Step 2.2: Combine CNN with graph convolutional network, and use graph convolutional network, i.e. GCN, to process the topological structure of spatial layout graph and capture the connection relationship between spatial units; Step 2.3: Input the preprocessed spatial layout map into SSP-CNN, i.e., spatial structure-aware convolutional neural network, to extract multi-level spatial feature maps; Step 2.4: Based on the pre-trained BERT model, adjust it using professional corpus in the field of medical architecture; Step 2.5: In semantic encoding, pay more attention to key demand words and use the attention mechanism to highlight important information; Step 2.6: Input the preprocessed natural language requirements into the adaptive bidirectional encoder representation model to obtain a high-dimensional requirement semantic encoding feature vector, which contains functional requirements, constraints and priority information; Step 2.7: When fusing spatial features and semantic features, establish a bidirectional attention mechanism so that the information of the two modalities can influence and complement each other; Step 2.8: Fusion is performed at different feature levels, taking into account both high-level abstract features and low-level detail information; Step 2.9: Input the spatial feature map and the requirement semantic encoding feature vector into the multimodal interactive attention fusion network to generate a fusion feature identifier and establish the association between spatial layout and functional requirements.

4. The auxiliary hospital space design method based on building energy consumption model according to claim 1 is characterized in that: Step 3 is as follows: Step 3.1: Introduce conditional information into the generator and discriminator so that the generation process is guided by the requirements; Step 3.2: The generator outputs multi-scale layout solutions, and the discriminator evaluates the rationality of the layout at different scales; Step 3.3: Prepare training data, including excellent hospital design plans and their corresponding functional requirements descriptions, and train the conditional generative adversarial network; Step 3.4: Use the real layout and condition information to train the discriminator so that it can distinguish between the real and generated layouts and determine whether the layout meets the conditions; Step 3.5: Train the generator to generate layout solutions that are difficult for the discriminator to distinguish and meet the requirements of conditional information; Step 3.6: Re-input the generated layout solution into the feature extractor and fusion network to obtain a new fusion feature representation; Step 3.7: Calculate the difference between the initial fused feature representation and the regenerated fused feature representation as the cycle consistency loss; Step 3.8: Input the user's fused feature representation into the trained generator to generate a preliminary spatial layout plan; Step 3.9: Use multi-objective optimization strategy to optimize the solution; Step 3.10: Output the optimized spatial layout plan, including detailed functional area division and energy consumption assessment data.

5. The auxiliary hospital space design method based on building energy consumption model according to claim 1 is characterized in that: Step 4 is as follows: Step 4.1: Define multiple evaluation indicators, each of which uses a weighted scoring mechanism; Step 4.1: Adjust the weights of each evaluation criterion through user feedback or simulation calculations, so that the evaluation process can adapt to different user needs; Step 4.2: Use CEM to comprehensively evaluate multiple candidate solutions and assign scores to each solution based on the evaluation criteria; Step 4.3: Compare the assessment results with user needs and identify the spatial layout solution that meets the best standards; Step 4.4: Use a multi-objective optimization algorithm to obtain multiple Pareto optimal solutions, and select the solution that best suits user needs through a decision support system; Step 4.5: Adjust the decision criteria based on user feedback and automatically recommend the best solution.

6. The auxiliary hospital space design method based on building energy consumption model according to claim 1 is characterized in that: Step 5 is as follows: Step 5.1: Use the generative adversarial network combined with AutoCAD automation tools to convert the final selected spatial layout plan into detailed architectural design drawings; Step 5.2: The generated design drawing includes multi-level design documents; Step 5.3: Use the deep learning model to simulate the construction process of the hospital building, simulating the construction difficulty, time schedule and cost; Step 5.4: Optimize the construction process through simulation results, provide the best construction plan, and reduce uncertainty and risks in construction; Step 5.5: During the construction implementation process, collect feedback information from actual use, automatically adjust the design plan, and perform iterative optimization; Step 5.1: Based on user feedback, automatically adjust the layout plan to achieve dynamic optimization.

7. A device for assisting hospital space design method based on building energy consumption model, applicable to the assisting hospital space design method based on building energy consumption model as claimed in any one of claims 1 to 6, characterized in that: It includes input module, cross-modal feature extraction module, spatial saliency and demand guidance module, energy efficiency analysis and optimization module, decision support and layout generation module and construction simulation and feedback adjustment module.

8. The device for assisting hospital space design method based on building energy consumption model according to claim 7 is characterized in that: The input module is used to obtain the hospital design requirements input by the user through natural language processing technology; the user inputs the hospital space boundary map through the graphical interface or uploads the existing space layout using a scanned image; The cross-modal feature extraction module uses a convolutional neural network to perform deep learning feature extraction on the input apartment boundary map to generate a shallow feature map of the spatial layout; it performs semantic encoding on the user's natural language description and converts the demand into a high-dimensional semantic feature vector through the BERT or Transformer model; The spatial saliency and demand guidance module focuses on the spatially significant areas through multi-scale convolutional pooling and attention mechanism; interactively encodes the semantic features of user needs and spatial saliency features to generate demand-guided spatial layout solutions; The energy efficiency analysis and optimization module evaluates the energy efficiency of each design scheme and calculates the energy consumption of each item based on the building energy efficiency model, namely BEM. It optimizes the scheme according to the energy efficiency simulation results and user needs through intelligent optimization algorithms. The decision support and layout generation module conducts a comprehensive evaluation based on multi-dimensional evaluation criteria and selects the design solution that meets the requirements; through the technology combining GAN and AutoCAD, it automatically generates architectural design drawings; The construction simulation and feedback adjustment module uses deep learning and building information modeling, or BIM, to simulate the construction process and simulate resource scheduling, time management, and risk assessment during the construction process; After the design plan is implemented, the actual use of the building is monitored in real time through the Internet of Things technology, feedback data is collected and the design plan is adjusted to form a closed-loop optimization.

Images