City block form optimization design decision-making method and system based on generation of large model

Abstract

The invention provides an urban block form optimization design decision-making method and system based on generation of a large model. The method comprises the steps that multi-mode green performance data influencing local heat island intensity and wind environment performance under the city block scale are collected and preprocessed; converting the preprocessed data into RDF triples according to entities, relationships and events, and constructing a multi-modal knowledge graph; performing modular clustering on knowledge in the multi-modal knowledge graph, obtaining a structured sub-graph, improving the knowledge retrieval efficiency based on the structured sub-graph, and obtaining a building green performance data vector database for establishing an external knowledge base enhanced large language model; and providing guidance and receiving feedback for a machine learning model for executing urban block form optimization design through a knowledge representation enhanced large language model, and forming a large and small model collaborative end-cloud collaborative architecture. According to the method and the system, the multi-modal data processing capability of a traditional single machine learning method can be improved, and the efficiency and the precision of optimization design decision making are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of green building performance optimization design, and in particular relates to a method and system for optimizing the design of urban block forms based on generating a large model. Background Art

[0002] Significant progress has been made in the research of green building performance simulation based on physical models and machine learning models at home and abroad. Most of the existing green building performance simulations are based on the assumption of uniform mixing, while the microclimate at the scale of urban blocks will bring non-uniform boundary conditions. Therefore, improving the accuracy of green building performance optimization requires in-depth research on non-uniform environment prediction and microclimate coupling through multimodal data.

[0003] In recent years, large language models (LLMs) have demonstrated excellent multimodal understanding and generation capabilities. These models are able to handle inputs and outputs of multiple modalities, including text, images, videos, audio, etc., including understanding and generating multimodal content, modality-specific generation, and imitating human modal conversion. These capabilities are driving the development of multimodal large language models. However, when using LLMs, researchers have also discovered the problem of hallucination. Existing studies have shown that by injecting knowledge graph information into LLMs in different forms, the model's hallucination problem can be significantly improved. The core idea of ​​this method is to use the knowledge graph to retrieve triple knowledge related to the entities explicitly mentioned in the input question, thereby providing LLMs with rich factual information.

[0004] For green building performance simulation, machine learning-based building performance simulation has significant advantages in improving speed and reducing hardware requirements, but its generalization is insufficient, it is difficult to process multimodal data, and it is highly professional and difficult to popularize. Therefore, combining generative design and big language technology can improve the accuracy of optimization design and promote efficient decision-making in building green performance optimization design. Summary of the invention

[0005] The purpose of the present invention is to solve the problems in the prior art and to propose a method and system for optimizing the design of urban block morphology based on generating a large model.

[0006] The present invention is implemented by the following technical scheme. The present invention proposes a method for optimizing the design of urban block morphology based on generating a large model. The method comprises the following steps:

[0007] S1. Collect and pre-process multi-modal green performance data that affect local heat island intensity and wind environment performance at the urban block scale;

[0008] The step S1 comprises the following steps:

[0009] S11. Collect multimodal green performance data;

[0010] S12, generating an embedded representation that fuses multimodal information;

[0011] S13, extracting information from the data;

[0012] S2. Convert the pre-processed multimodal green performance data into RDF triples according to entities, relationships, and events, and construct a multimodal knowledge graph;

[0013] The step S2 comprises the following steps:

[0014] S21, converting the extracted information into RDF triples;

[0015] S22, constructing a multimodal knowledge graph based on RDF triples;

[0016] S3, modularly clustering the knowledge in the multimodal knowledge graph and obtaining a structured subgraph, improving the knowledge retrieval efficiency based on the structured subgraph, and obtaining a building green performance data vector database for establishing an external knowledge base to enhance the large language model;

[0017] The step S3 comprises the following steps:

[0018] S31. Modularize the knowledge in the knowledge graph and obtain a structured subgraph;

[0019] S32, improving knowledge retrieval efficiency based on structured subgraphs;

[0020] S33, obtaining a building green performance data vector database;

[0021] S4, using the knowledge representation to enhance the large language model to provide guidance and receive feedback for the machine learning model that performs the urban block morphology optimization design, thereby forming an end-cloud collaborative architecture that collaborates with large and small models;

[0022] The step S4 comprises the following steps:

[0023] S41: The cloud-side big language model receives the designer's optimization design task instructions and related information, thereby understanding the designer's requirements and providing simulation instructions to the client-side model.

[0024] S42. The client-side model combines the cloud-side instructions with the relevant engineering scenario model files and relevant urban block environmental parameters to perform an urban block morphology optimization simulation and provide simulation result sample data.

[0025] S43. The large language model analyzes the simulation result sample data and provides decision-making guidance.

[0026] Furthermore, in step S11, a multimodal green performance database affecting the local heat island intensity and wind environment performance at the urban block scale is constructed through dynamic acquisition of CIM, BIM or IoT integrated platform data, and the data modality type and corresponding data format requirements in the data analysis process are determined;

[0027] In the step S12, the CLIP or MMV cross-modal alignment algorithm and the multimodal Transformer are used to process the green performance data of buildings in different modes, and these data are uniformly converted into high-dimensional vectors that retain their multimodal data characteristics, and embedded in a shared vector space to achieve information fusion and alignment, retain and strengthen the association between the green performance data of buildings in different modes, so as to assist in the understanding of the semantic relationship of the multimodal data, so that the model can simultaneously consider the information of multiple modes when processing multimodal tasks;

[0028] In the step S13, the information extraction includes entity extraction, relationship extraction and event extraction; the entity extraction refers to applying a named entity recognition method based on a BiLSTM-CRF deep learning model, using context information to extract specific building green performance objects or concepts from an unstructured multi-scale building green performance dataset; the relationship extraction refers to using a relational classification model or a dependency syntax analysis method to identify the relationship between the above-mentioned building green performance entities; the event extraction is detected by an ACE event extraction framework, and identifies building green performance event trigger words and related information based on a deep learning model.

[0029] Furthermore, in the step S21, the building green performance data is converted into RDF triples of "entity-relationship-entity" through Python's RDF toolkit RDFLib, and the triples are stored in a graph database for subsequent processing and query operations;

[0030] In step S22, the extracted entities are aligned with the existing entities in the knowledge base, a relationship graph between entities is constructed, and the entities are linked using a graph matching algorithm; for records pointing to a unified entity in the knowledge base, the distance calculations of entities and relationships embedded in the vector space are merged using a clustering algorithm or a DeepWalk algorithm; the relationship similarity is calculated using an embedding model, and the relationships between the same entities in different data sources are integrated using a graph neural network; the fused data is enhanced through reasoning to improve the accuracy of information extraction and the reasoning ability of the knowledge graph.

[0031] Furthermore, in step S31, the community detection algorithm is used to identify the interrelated node groups, and the knowledge graph is divided into different communities; the graph embedding algorithm is used to identify and add missing relationships or entities, so as to enhance the integrity of the knowledge graph and assist knowledge reasoning;

[0032] In step S32, a hierarchical clustering method is used to extract abstracts at multi-dimensional scales of cities, blocks, buildings, and spaces, and a hierarchical analysis of communities is performed; structured subgraphs and their abstracts are converted into high-dimensional vector representations; high-dimensional vector retrieval is assisted by FAISS or HNSW algorithms to improve the efficiency of machine learning tasks;

[0033] In the step S33, an enhanced foundation model is constructed by obtaining the building green performance data vector database.

[0034] Furthermore, in step S41, the large language model, as a vertical architectural question-and-answer assistant, opens the intelligent question-and-answer service function to the designer, and the input end receives the designer's optimization design task instructions and the input of related multimodal data, and then understands the designer's needs and converts them into simulation instructions to provide to the end-side model;

[0035] In step S42, the client-side model is a machine learning model. Building performance simulation based on machine learning has significant advantages in improving speed and reducing hardware requirements.

[0036] In step S43, through the end-cloud collaborative architecture, the big language model on the cloud side combines the natural language instructions given by the designer with related information to generate optimization design simulation instructions, and the end-side machine learning model executes the urban block morphology optimization simulation to give simulation result sample data, which is fed back to the big language model on the cloud side for further analysis and decision guidance.

[0037] The present invention also proposes an urban block morphology optimization design decision system based on generating a large model, the system comprising:

[0038] Acquisition module: Integrate remote sensing satellites and drone platforms to monitor urban thermal environment from two scales, collect urban surface temperature data and urban physical environment information, study the intensity and distribution characteristics of urban heat island effects in different areas, and analyze the impact of human activity intensity;

[0039] Retrieval module: Based on the retrieval task, the designer's retrieval content is converted into a prompt word project, the retrieval results are limited to the vertical field of architecture, and the retrieval results are determined through a large language model;

[0040] Decision-making module: The big language model inputs the search results into the machine learning model to further complete the optimization simulation of urban block forms. The machine learning model can feed back the simulation results to the big language model for further decision-making. The end-cloud collaboration assists in the optimization design.

[0041] Interactive module: Establish a digital twin model of urban block morphology, provide a visual expression of building layout and building height under simulation results, and intuitively display the impact of building morphology on urban heat island intensity and block microenvironment climate.

[0042] Furthermore, in the acquisition module, at the city scale, the urban thermal environment is monitored based on remote sensing satellites equipped with MODIS, and urban surface temperature data and a series of remote sensing satellite images are collected; at the block scale, based on the DJI FlyCart 30 medium-sized multi-rotor drone platform, infrared thermal imaging sensors, temperature and humidity sensors, and ultrasonic wind speed sensors are integrated to further collect temperature difference, air humidity, air velocity, air flow, urban surface roughness, urban underlying surface morphology, and urban building morphology physical environment information to analyze the urban block microenvironment climate;

[0043] The designer's retrieval content includes natural language text and image data, and the large language model retrieval results are generated as a block morphology design plan to improve the local heat island intensity and wind environment performance, so as to further assist the machine learning model in performing optimization simulation tasks.

[0044] Furthermore, the decision module is an end-cloud collaborative architecture that combines an end-side machine learning model with a cloud-side big language model. Combined with typical engineering scenarios for optimizing urban block morphology design, the big language model uses cloud computing power to transform the design decision-making model based on subjective experience into a computational design decision-making model under vertical big model knowledge fusion; the end-side machine learning model takes advantage of the low latency and real-time processing characteristics of the end-side to form a base model collaborative architecture of "intelligent decision-making in the cloud and real-time response at the edge".

[0045] The present invention also proposes an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for optimizing the design and decision-making of urban block morphology based on generating a large model are implemented.

[0046] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the urban block morphology optimization design decision method based on generating a large model.

[0047] Beneficial effects of the present invention:

[0048] The present invention provides a method and system for optimizing the design and decision-making of urban block morphology based on generating a large model. The method and system can improve the multimodal data processing capability of traditional single machine learning methods and enhance the efficiency and accuracy of optimizing design and decision-making. Combining generative design and large language model technology, a question-answering assistant in the vertical field of architecture is constructed, which is conducive to guiding efficient decision-making in optimizing the design of green performance of buildings and providing strong support for the rapid and efficient simulation of green building performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0050] Figure 1 The present invention is a flow chart of a method for optimizing design of urban block morphology based on generating a large model. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Combination Figure 1 The present invention proposes a method for optimizing the design of urban block morphology based on generating a large model, and the method comprises the following steps:

[0053] S1. Collect and pre-process multi-modal green performance data that affect local heat island intensity and wind environment performance at the urban block scale;

[0054] The step S1 comprises the following steps:

[0055] S11. Collect multimodal green performance data;

[0056] S12, generating an embedded representation that fuses multimodal information;

[0057] S13, extracting information from the data;

[0058] In the step S11, a multimodal green performance database affecting the local heat island intensity and wind environment performance at the urban block scale is constructed by dynamically acquiring data from the CIM, BIM or IoT integrated platform, and the data modality type and corresponding data format requirements in the data analysis process are determined;

[0059] In the step S12, the CLIP or MMV cross-modal alignment algorithm and the multimodal Transformer are used to process the green performance data of buildings in different modes, and these data are uniformly converted into high-dimensional vectors that retain their multimodal data characteristics, and embedded in a shared vector space to achieve information fusion and alignment, and the association between the green performance data of buildings in different modes, such as the association between images and text descriptions, is retained and strengthened to assist in understanding the semantic relationship of multimodal data, so that the model can simultaneously consider information from multiple modes when processing multimodal tasks;

[0060] In the step S13, the information extraction includes entity extraction, relationship extraction and event extraction; the entity extraction refers to the application of named entity recognition (NER) method based on BiLSTM-CRF deep learning model, using context information to extract specific building green performance objects or concepts from unstructured multi-scale building green performance data set; the relationship extraction refers to the use of OpenIE, BERT and other relational classification models or dependency syntax analysis methods to identify the relationship between the above-mentioned building green performance entities; the event extraction usually involves the spatiotemporal dimension parameters of multiple building green performance entities, which are detected by ACE event extraction framework, and based on BiLSTM-CRF, BERT and other deep learning models, identify building green performance event trigger words and related information.

[0061] S2. The pre-processed multimodal green performance data is converted into RDF triples according to entities, relationships, and events to construct a multimodal knowledge graph. Specifically: the pre-processed multimodal green performance data is integrated and associated with multi-source heterogeneous data using knowledge graph technology to form a structured knowledge system, and the urban form, climate background, geographical location, and human activity intensity are comprehensively modeled to construct a multimodal knowledge graph of urban heat island intensity and wind environment influencing factors.

[0062] The step S2 comprises the following steps:

[0063] S21, converting the extracted information into RDF triples;

[0064] S22. Constructing a multimodal knowledge graph based on RDF triples;

[0065] In the step S21, the building green performance data is converted into RDF triples of "entity-relationship-entity" through Python's RDF toolkit RDFLib, and the triples are stored in a graph database for subsequent processing and query operations;

[0066] In the step S22, the extracted entities are aligned with the existing entities in the knowledge base to construct a relationship graph between the entities, and the entities are linked using graph matching algorithms such as PageRank and Random Walk; for the records pointing to the same entity in the knowledge base, the distance calculations of the entities and relationships embedded in the vector space are merged using clustering algorithms such as K-means and DBSCAN or the DeepWalk algorithm; relationship similarity calculations are performed using embedding models such as Word2Vec and GloVe, and the relationships between the same entities from different data sources are integrated using graph neural networks such as GCN and RGCNs; the fused data is enhanced through reasoning to improve the accuracy of information extraction and the reasoning ability of the knowledge graph.

[0067] S3, modularly clustering the knowledge in the multimodal knowledge graph and obtaining a structured subgraph, improving the knowledge retrieval efficiency based on the structured subgraph, obtaining a building green performance data vector database for establishing an external knowledge base to enhance the large language model; obtaining the designer's design information for the urban block morphology optimization; based on the design information, using the large language model in combination with the multimodal knowledge graph to generate highly interpretable analysis result information to assist the machine learning model in simulating the optimization of the urban block morphology;

[0068] The step S3 comprises the following steps:

[0069] S31. Modularize the knowledge in the knowledge graph and obtain a structured subgraph;

[0070] S32, improving knowledge retrieval efficiency based on structured subgraphs;

[0071] S33, obtaining a building green performance data vector database;

[0072] In step S31, the community detection algorithm is used to identify the interrelated node groups and divide the knowledge graph into different communities; the graph embedding algorithm is used to identify and add missing relationships or entities to enhance the integrity of the knowledge graph and assist knowledge reasoning;

[0073] In step S32, hierarchical clustering methods such as hierarchical clustering and k-means clustering are used to extract abstracts at multi-dimensional scales of cities, blocks, buildings, and spaces, and to perform hierarchical analysis of communities; structured subgraphs and their abstracts are converted into high-dimensional vector representations through algorithms such as Node2Vec, DeepWalk, and GraphSAGE; high-dimensional vector retrieval is assisted by FAISS or HNSW algorithms to improve the efficiency of machine learning tasks such as clustering and retrieval;

[0074] In the step S33, an enhanced foundation model is constructed by obtaining the building green performance data vector database.

[0075] S4, using the knowledge representation to enhance the large language model to provide guidance and receive feedback for the machine learning model that performs the urban block morphology optimization design, thereby forming an end-cloud collaborative architecture that collaborates with large and small models;

[0076] The step S4 comprises the following steps:

[0077] S41: The cloud-side large language model receives the designer's optimization design task instructions and related inputs, understands user requirements, and provides simulation instructions to the client-side model.

[0078] S42. The client-side model combines the cloud-side instructions with the relevant engineering scenario model files and relevant urban block environmental parameters to perform an urban block morphology optimization simulation and provide simulation result sample data.

[0079] S43. The large language model analyzes the simulation result sample data and provides decision-making guidance.

[0080] In step S41, the large language model, as a vertical architectural question-and-answer assistant, opens the intelligent question-and-answer service function to the designer. The input end receives the designer's optimization design task instructions and the input of related multimodal data, and then understands the designer's needs and converts them into simulation instructions to provide to the end-side model.

[0081] In step S42, the client-side model is a machine learning model. Building performance simulation based on machine learning has significant advantages in improving speed and reducing hardware requirements.

[0082] In step S43, through the end-cloud collaborative architecture, the big language model on the cloud side combines the natural language instructions given by the designer with related information to generate optimization design simulation instructions, and the end-side machine learning model executes the urban block morphology optimization simulation to give simulation result sample data, which is fed back to the big language model on the cloud side for further analysis and decision guidance.

[0083] The present invention also proposes an urban block morphology optimization design decision system based on generating a large model, the system comprising:

[0084] Acquisition module: Integrate remote sensing satellites and drone platforms to monitor urban thermal environment from two scales, collect urban surface temperature data and urban physical environment information, study the intensity and distribution characteristics of urban heat island effects in different areas, and analyze the impact of human activity intensity;

[0085] Retrieval module: Based on the retrieval task, the designer's retrieval content is converted into a prompt word project, the retrieval results are limited to the vertical field of architecture, and the retrieval results are determined through a large language model;

[0086] Decision-making module: The big language model inputs the search results into the machine learning model to further complete the optimization simulation of urban block forms. The machine learning model can feed back the simulation results to the big language model for further decision-making. The end-cloud collaboration assists in the optimization design.

[0087] Interactive module: Establish a digital twin model of urban block morphology, provide a visual expression of building layout and building height under simulation results, and intuitively display the impact of building morphology on urban heat island intensity and block microenvironment climate.

[0088] The multimodal green performance data includes numerical values ​​such as temperature difference, wind speed, building density, building height, texts such as government documents, engineering documents, specifications and standards, and images such as low-altitude aerial photos of urban blocks; the data acquisition method includes dynamic acquisition of data from integrated platforms such as CIM, BIM, and IoT; the data preprocessing method includes using cross-modal alignment algorithms such as CLIP, MMV, and multimodal Transformer processing to uniformly convert the data into high-dimensional vectors that retain their multimodal data characteristics, and then extract information through entity extraction, relationship extraction, and event extraction to retain the entity information and relationship information required for knowledge graph construction.

[0089] In the acquisition module, at the city scale, the urban thermal environment is monitored based on remote sensing satellites equipped with MODIS (Moderate Resolution Imaging Spectroradiometer), and urban surface temperature data and a series of remote sensing satellite images are collected; at the block scale, based on the DJI FlyCart 30 medium-sized multi-rotor drone platform integrated with infrared thermal imaging sensors, temperature and humidity sensors, and ultrasonic wind speed sensors, further collect temperature difference, air humidity, air velocity, air flow, urban surface roughness, urban underlying surface morphology and urban building morphology physical environment information to analyze the microenvironment climate of urban blocks; three-dimensional reconstruction technology can also be used to establish a digital twin model of urban block morphology to assist in decision-making visualization.

[0090] The designer's retrieval content includes natural language text and image data, such as urban environmental information, urban and architectural form spatial data, and low-altitude aerial photos of urban blocks collected by the acquisition module, including design task texts, government documents, and specifications and standards; the large language model retrieval results are generated as block form design plans for improving local heat island intensity and wind environment performance, including design parameters such as building height, building form, and building location, to further assist the machine learning model in performing optimization simulation tasks.

[0091] The decision module is an end-cloud collaborative architecture that combines an end-side machine learning model with a cloud-side big language model. Combined with typical engineering scenarios for optimizing urban block morphology, the big language model uses cloud computing power to transform the design decision-making model based on subjective experience into a computational design decision-making model under vertical big model knowledge fusion; the end-side machine learning model takes advantage of the low latency and real-time processing characteristics of the end-side to form a base model collaborative architecture of "intelligent decision-making in the cloud and real-time response at the edge".

[0092] In the interactive module, the urban block digital twin system can improve the visualization and intelligence of the optimized design. It uses drone three-dimensional reconstruction technology to display high-precision 3D models of urban blocks through a digital sand table, simulate the architectural form and layout of urban blocks, and further analyze the local heat island intensity and wind environment climate of urban blocks to achieve visual intelligent decision-making support.

[0093] The present invention also proposes an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for optimizing the design and decision-making of urban block morphology based on generating a large model are implemented.

[0094] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the urban block morphology optimization design decision method based on generating a large model.

[0095] The memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus RAM (DRRAM). It should be noted that the memory of the method described in the present invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0096] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a high-density digital video disc (DVD)), or a semiconductor medium (eg, a solid state disc (SSD)).

[0097] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software. The steps of the method disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software modules in a processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it is not described in detail here.

[0098] It should be noted that the processor in the embodiment of the present application can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor can be combined and performed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0099] The above is a detailed introduction to the urban block morphology optimization design decision method and system based on generating a large model proposed in the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for optimizing the design of urban block morphology based on generating a large model, characterized in that: The method comprises the following steps: S1. Collect and pre-process multi-modal green performance data that affect local heat island intensity and wind environment performance at the urban block scale; The step S1 comprises the following steps: S11. Collect multimodal green performance data; S12, generating an embedded representation that fuses multimodal information; S13, extracting information from the data; S2. Convert the pre-processed multimodal green performance data into RDF triples according to entities, relationships, and events, and construct a multimodal knowledge graph; The step S2 comprises the following steps: S21, converting the extracted information into RDF triples; S22, constructing a multimodal knowledge graph based on RDF triples; S3, modularly clustering the knowledge in the multimodal knowledge graph and obtaining a structured subgraph, improving the knowledge retrieval efficiency based on the structured subgraph, and obtaining a building green performance data vector database for establishing an external knowledge base to enhance the large language model; The step S3 comprises the following steps: S31. Modularize the knowledge in the knowledge graph and obtain a structured subgraph; S32, improving knowledge retrieval efficiency based on structured subgraphs; S33, obtaining a building green performance data vector database; S4, using the knowledge representation to enhance the large language model to provide guidance and receive feedback for the machine learning model that performs the urban block morphology optimization design, thereby forming an end-cloud collaborative architecture that collaborates with large and small models; The step S4 comprises the following steps: S41: The cloud-side large language model receives the designer's optimization design task instructions and related information, thereby understanding the designer's requirements and providing simulation instructions to the client-side model. S42. The client-side model combines the cloud-side instructions with the relevant engineering scenario model files and relevant urban block environmental parameters to perform an urban block morphology optimization simulation and provide simulation result sample data. S43. The large language model analyzes the simulation result sample data and provides decision-making guidance.

2. The method according to claim 1, characterized in that In the step S11, a multimodal green performance database affecting the local heat island intensity and wind environment performance at the urban block scale is constructed by dynamically acquiring data from the CIM, BIM or IoT integrated platform, and the data modality type and corresponding data format requirements in the data analysis process are determined; In the step S12, the CLIP or MMV cross-modal alignment algorithm and the multimodal Transformer are used to process the green performance data of buildings in different modes, and these data are uniformly converted into high-dimensional vectors that retain their multimodal data characteristics, and embedded in a shared vector space to achieve information fusion and alignment, retain and strengthen the association between the green performance data of buildings in different modes, so as to assist in the understanding of the semantic relationship of the multimodal data, so that the model can simultaneously consider the information of multiple modes when processing multimodal tasks; In the step S13, the information extraction includes entity extraction, relationship extraction and event extraction; the entity extraction refers to applying a named entity recognition method based on a BiLSTM-CRF deep learning model, using context information, to extract specific building green performance objects or concepts from an unstructured multi-scale building green performance data set; the relationship extraction refers to using a relationship classification model or a dependency syntax analysis method to identify the relationship between the above-mentioned building green performance entities; The event extraction is detected through the ACE event extraction framework, and the building green performance event trigger words and related information are identified based on the deep learning model.

3. The method according to claim 1, characterized in that In the step S21, the building green performance data is converted into RDF triples of "entity-relationship-entity" through Python's RDF toolkit RDFLib, and the triples are stored in a graph database for subsequent processing and query operations; In the step S22, the extracted entities are aligned with the existing entities in the knowledge base to construct a relationship graph between the entities, and the entities are linked using a graph matching algorithm; for the records pointing to the same entity in the knowledge base, the distance calculation of the entities and relationships embedded in the vector space is merged by a clustering algorithm or a DeepWalk algorithm; the relationship similarity is calculated using an embedding model, and the relationship between the same entities in different data sources is integrated using a graph neural network; The fused data is enhanced through reasoning to improve the accuracy of information extraction and the reasoning ability of the knowledge graph.

4. The method according to claim 1, characterized in that: In step S31, the community detection algorithm is used to identify the interrelated node groups and divide the knowledge graph into different communities; the graph embedding algorithm is used to identify and add missing relationships or entities to enhance the integrity of the knowledge graph and assist knowledge reasoning; In step S32, a hierarchical clustering method is used to extract abstracts at multi-dimensional scales of cities, blocks, buildings, and spaces, and a hierarchical analysis of communities is performed; structured subgraphs and their abstracts are converted into high-dimensional vector representations; high-dimensional vector retrieval is assisted by FAISS or HNSW algorithms to improve the efficiency of machine learning tasks; In the step S33, an enhanced foundation model is constructed by using the obtained building green performance data vector database.

5. The method according to claim 1, characterized in that In step S41, the large language model, as a vertical architectural question-and-answer assistant, opens the intelligent question-and-answer service function to the designer. The input end receives the designer's optimization design task instructions and the input of related multimodal data, and then understands the designer's needs and converts them into simulation instructions to provide to the end-side model. In step S42, the client-side model is a machine learning model. Building performance simulation based on machine learning has significant advantages in improving speed and reducing hardware requirements. In step S43, through the end-cloud collaborative architecture, the big language model on the cloud side combines the natural language instructions given by the designer with related information to generate optimization design simulation instructions, and the end-side machine learning model executes the urban block morphology optimization simulation to give simulation result sample data, which is fed back to the big language model on the cloud side for further analysis and decision guidance.

6. A city block morphology optimization design decision system based on generating a large model, characterized in that: The system comprises: Acquisition module: Integrate remote sensing satellites and drone platforms to monitor urban thermal environment from two scales, collect urban surface temperature data and urban physical environment information, study the intensity and distribution characteristics of urban heat island effects in different areas, and analyze the impact of human activity intensity; Retrieval module: Based on the retrieval task, the designer's retrieval content is converted into a prompt word project, the retrieval results are limited to the vertical field of architecture, and the retrieval results are determined through a large language model; Decision-making module: The big language model inputs the search results into the machine learning model to further complete the optimization simulation of urban block forms. The machine learning model can feed back the simulation results to the big language model for further decision-making. The end-cloud collaboration assists in the optimization design. Interactive module: Establish a digital twin model of urban block morphology, provide a visual expression of building layout and building height under simulation results, and intuitively display the impact of building morphology on urban heat island intensity and block microenvironment climate.

7. The system according to claim 6, characterized in that In the acquisition module, at the city scale, the urban thermal environment is monitored based on remote sensing satellites equipped with MODIS, and urban surface temperature data and a series of remote sensing satellite images are collected; at the block scale, based on the DJI FlyCart 30 medium-sized multi-rotor drone platform, infrared thermal imaging sensors, temperature and humidity sensors, and ultrasonic wind speed sensors are integrated to further collect temperature difference, air humidity, air velocity, air flow, urban surface roughness, urban underlying surface morphology, and urban building morphology physical environment information to analyze the urban block microenvironment climate; The designer's retrieval content includes natural language text and image data, and the large language model retrieval results are generated as a block morphology design plan to improve the local heat island intensity and wind environment performance, so as to further assist the machine learning model in performing optimization simulation tasks.

8. The system according to claim 6, characterized in that The decision module is an end-cloud collaborative architecture that combines an end-side machine learning model with a cloud-side big language model. Combined with typical engineering scenarios for optimizing urban block morphology, the big language model uses cloud computing power to transform the design decision-making model based on subjective experience into a computational design decision-making model under vertical big model knowledge fusion; the end-side machine learning model takes advantage of the low latency and real-time processing characteristics of the end-side to form a base model collaborative architecture of "intelligent decision-making in the cloud and real-time response at the edge".

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium for storing computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images