A three-dimensional form optimization method for cold region neighborhood building group based on explainable artificial intelligence

Abstract

This invention proposes a method for optimizing the 3D morphology of building complexes in cold-region neighborhoods based on interpretable artificial intelligence. This method integrates multi-objective optimization algorithms and interpretable artificial intelligence technology to achieve automatic iterative optimization and result interpretation of 3D morphology schemes for building complexes. Addressing the unique climatic characteristics of cold regions, this method automates the entire process from generation and optimization to result interpretation of 3D morphology schemes for neighborhood-scale building complexes, significantly improving the quality and practicality of design schemes and possessing significant practical application value.

Description

Technical Field

[0001] This invention relates to the fields of urban planning and design and artificial intelligence technology, and in particular to a method for optimizing the three-dimensional form of building complexes in cold-region neighborhoods based on interpretable artificial intelligence. This method integrates multi-objective optimization algorithms and interpretable artificial intelligence technology to achieve automatic iterative optimization and result interpretation of the three-dimensional form schemes for building complexes. Background Technology

[0002] Currently, the three-dimensional morphological design of neighborhood-scale buildings primarily relies on manual rule verification and the subjective experience of designers. Even after a design scheme meets rigid regulatory requirements, its refinement process still largely depends on the designer's experience and judgment, which typically focuses on a single objective and lacks a comprehensive consideration of multiple goals. Especially in cold regions, when neighborhood building complexes need to simultaneously address multiple requirements such as sunlight, ventilation, and energy conservation, human experience at the design stage can lead to deviations in the performance of the building complex during operation. Even when tools such as building performance simulation are used to assist in optimization during the design process, manual verification of the scheme's feasibility is still required, resulting in very low efficiency. Therefore, there is an urgent need to develop a transferable and reusable automated design and optimization process for cold-region neighborhood building complexes.

[0003] Furthermore, existing intelligent optimization processes lack interpretability. When intelligent optimization tools such as machine learning or genetic algorithms are introduced, the decision-making process becomes a "black box," making it difficult for designers to understand how the optimization results are generated or to determine which design parameters play a key role in the quality of the solution. This lack of interpretability reduces the credibility of the solution and hinders the application of design results in practical planning. Summary of the Invention

[0004] The purpose of this invention is to address the problems in existing technologies by proposing a three-dimensional morphological optimization method for cold-region neighborhood building complexes based on interpretable artificial intelligence. This method, tailored to the unique climatic characteristics of cold regions, automates the entire process from generation and optimization of three-dimensional morphological schemes for neighborhood-scale building complexes to result interpretation. It significantly improves the quality and practicality of design schemes and has significant practical application value.

[0005] This invention is achieved through the following technical solution: This invention proposes a method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence, the method comprising: Step 1: Establish a morphological prototype library and parameter system; based on the design requirements of urban block-scale building complexes in cold regions, pre-construct a three-dimensional morphological prototype model library, establish a corresponding morphological parameterization representation system, and statistically analyze parameter ranges to describe the morphological characteristics of the building complex. Step two: Determine the constraints and evaluation index system; transform the mandatory requirements of urban planning into hard constraints, and at the same time construct a soft performance index system that reflects the spatial quality and environmental performance of cold-region neighborhood building clusters, which together serve as subsequent optimization objectives. Step 3: Generate a batch of scheme combinations and construct an indicator data dataset; construct a large number of morphological schemes based on random sampling, and output the evaluation index results of each scheme through calculation and simulation, and summarize them to form a training dataset; Step 4: Build a prediction model; use the above dataset to train the prediction model to achieve rapid evaluation and prediction of the scheme's indicators, thereby accelerating the subsequent optimization process. Step 5: Multi-objective Pareto evolution optimization; The NSGA-III algorithm is used to iteratively optimize and search for multiple objectives under the premise of satisfying hard constraints, so as to obtain a set of Pareto optimal solutions. Step 6: Interpretable AI Attribution Analysis; Perform interpretable analysis on the obtained optimal solution set to identify the dominant design parameters and their impact on various objectives; Step 7: Output the optimization results and generate an explanatory report; finally, output the Pareto solution set obtained from the optimization and the corresponding indicators, and generate a report containing the impact analysis of key parameters for decision-making reference.

[0006] Furthermore, step one specifically includes: Step 1.1, Morphological Prototype Collection and Construction: Collect typical neighborhood building layouts and individual building morphological cases from cold-region cities, extract building layout patterns and building shape types; based on these case morphologies, construct a parametric 3D morphological prototype model library; Step 1.2: Defining and determining the morphological parameters and their ranges; Based on the morphological prototype model, determine the key parameters describing the morphology of the building complex and their physical meaning; The parameters include two parts: one is the parameters of individual buildings within the building complex; the other is the interrelationship between buildings within the building complex; Calculate and statistically analyze the parameter values ​​for each case, and combine the planning and design requirements of cold regions with engineering experience to set a reasonable range for each parameter. Step 1.3: Establishing a prototype library and parametric representation; Integrate the various morphological parameters defined above into the prototype model extracted in Step 1, and realize the parametric representation of the building complex morphology on the Grasshopper platform; By assigning and changing different parameter combination values, generate different building complex morphological schemes from the prototype library; Finally, establish a morphological prototype library covering multiple layout types and parameter combinations, providing a complete morphological base and parameter support for subsequent automatic scheme generation.

[0007] Furthermore, in step two, the hard constraints are building density, floor area ratio, or building spacing; the soft performance index system is divided into two categories in the context of cold-region design: spatial quality and environmental performance. Spatial quality includes ventilation corridor width, ventilation corridor connectivity, and street continuity; environmental performance includes heating energy consumption intensity during the heating season, effective solar utilization rate in winter, and outdoor wind chill equivalent temperature.

[0008] Furthermore, step three specifically includes: Step 3.1: Automatic scheme generation; using the morphological prototype library established in Step 1, the determined parameter range, and the constructed parametric modeling process, a large number of building group scheme sets are generated in batches through random sampling. Step 3.2, Scheme Indicator Calculation and Simulation: For each automatically generated morphological design scheme, compliance verification is first performed. The hard constraint indicators of the scheme are calculated to see if they meet the set limits. Non-compliant schemes are marked as invalid. Then, for compliant schemes, code is written to calculate the spatial quality indicators. At the same time, based on the Grasshopper platform's Ladybug, Honeybee, and Butterfly, environmental performance indicators are simulated to obtain the indicator results. Step 3.3: Form a training set of scheme data; summarize the parameters and corresponding index calculation results of all candidate schemes to form a complete scheme dataset mapping; this dataset is a training sample set, in which each data point contains the parameter combination of the scheme and the results of various indexes of the scheme.

[0009] Furthermore, step four specifically includes: Step 4.1: Select and build a prediction model; select multiple artificial intelligence algorithms based on the characteristics and scale of the dataset to build a prediction model; the algorithms include MLP, RF, SVR, Catboost, LightGRM, and XGBoost; randomly divide the dataset into training and validation sets, train the selected model to enable it to quickly predict the corresponding index values ​​based on the input design parameters; during the training process, use the grid search method to adjust the model hyperparameters to improve prediction accuracy, and evaluate the error level of the model through the validation set to ensure that each prediction model has sufficient accuracy; Step 4.2, Multi-model Comparison and Validation: After the prediction model training converges and achieves the expected accuracy on the validation set, conduct a multi-algorithm comparison; using R... 2 The accuracy of the results is verified using three metrics: RMSE, MAE, and RMSE, and the optimal model is selected accordingly.

[0010] Furthermore, step five specifically includes: Step 5.1: Multi-objective optimization initialization; Select the NSGA-Ⅲ algorithm and set the parameters. Use the scheme set as the initial population. Each individual is encoded as a set of building group morphology parameters, and the corresponding index is obtained by the trained prediction model for rapid prediction. Step 5.2, Evolutionary Iteration Process: In each generation of optimization iteration, individuals in the population are selected, crossovered, and mutated based on multi-objective fitness to generate new candidate solutions. For each newly generated solution individual, its various index values ​​are first quickly evaluated through a prediction model, and then its compliance with hard constraints is checked: if an individual violates the hard constraints, the iteration is restarted; compliant individuals participate in Pareto sorting and crowding calculation operations according to the requirements of the optimization algorithm to determine their survival probability in the population. The non-dominated sorting mechanism is used to retain individuals that perform well in multi-objectives and eliminate inferior individuals. Step 5.3, Convergence Judgment and Termination: Repeat the evolutionary iteration process and continuously update the population; the iteration ends when the preset termination condition is reached; the termination condition is set according to the optimization requirements. After the iteration ends, all non-dominated solutions in the final population are extracted as the result scheme set of this optimization; each scheme in the obtained Pareto front scheme set has satisfied the hard constraints and achieved different trade-off combinations between spatial quality and environmental performance indicators.

[0011] Furthermore, step six specifically includes: We select interpretable methods and obtain the importance of global features; for the Pareto solution set obtained by optimization, we use interpretable artificial intelligence technology to perform attribution analysis on the solution performance; we use SHAP value analysis method to calculate the contribution value of each morphological parameter to each target index, and use PDP diagram to calculate the threshold and nonlinear response of morphological parameters to each target value. Through these mechanisms, we use interpretive methods to help designers understand the causes of optimization solutions.

[0012] Furthermore, step seven specifically includes: Step 7.1: Output the optimized scheme set and indicators; Output the final Pareto front scheme set and list the target indicator values ​​for each scheme; Make it easy for designers to view and retrieve the schemes through lists, tables or 3D model scheme integration files; For each scheme, clearly give its key parameter configuration and the achieved spatial quality index value and environmental performance index value. Step 7.2: Generate an interpretable optimization report; Based on the analysis results, an interpretable report document containing key information about the optimization process is automatically generated; This report, combined with interpretable methods, displays the main parameters and their importance ranking, and summarizes the direction of the influence of design parameters on each indicator.

[0013] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence.

[0014] 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 method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence.

[0015] The beneficial effects of this invention are: 1. Multi-objective integrated optimization based on the special climate conditions of cold regions: Under the premise of considering the compliance of building complexes in cold regions with constraints, the spatial quality and environmental performance are improved, providing more scientific design guidance for urban planning in cold regions.

[0016] 2. Improve design efficiency and credibility: Evolutionary algorithms are used to efficiently search for and optimize solutions, significantly improving design efficiency; interpretable artificial intelligence is introduced to attribute and explain the results, making the optimization process transparent and auditable, thus enhancing the credibility and persuasiveness of the solutions. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0018] Figure 1 This is a flowchart of a three-dimensional morphological optimization method for cold-region neighborhood building complexes based on interpretable artificial intelligence, as described in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Specifically, in combination Figure 1 This invention proposes a method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence, the method comprising: Step 1: Establish a morphological prototype library and parameter system; based on the design requirements of urban block-scale building complexes in cold regions, pre-construct a three-dimensional morphological prototype model library, establish a corresponding morphological parameterization representation system, and statistically analyze parameter ranges to describe the morphological characteristics of the building complex. Step two: Determine the constraints and evaluation index system; transform the mandatory requirements of urban planning into hard constraints, and at the same time construct a soft performance index system that reflects the spatial quality and environmental performance of cold-region neighborhood building clusters, which together serve as subsequent optimization objectives. Step 3: Generate a batch of scheme combinations and construct an indicator data dataset; construct a large number of morphological schemes based on random sampling, and output the evaluation index results of each scheme through calculation and simulation, and summarize them to form a training dataset; Step 4: Build a prediction model; use the above dataset to train the prediction model to achieve rapid evaluation and prediction of the scheme's indicators, thereby accelerating the subsequent optimization process. Step 5: Multi-objective Pareto evolution optimization; The NSGA-III algorithm is used to iteratively optimize and search for multiple objectives under the premise of satisfying hard constraints, so as to obtain a set of Pareto optimal solutions. Step 6: Interpretable AI Attribution Analysis; Perform interpretable analysis on the obtained optimal solution set to identify the dominant design parameters and their impact on various objectives; Step 7: Output the optimization results and generate an explanatory report; finally, output the Pareto solution set obtained from the optimization and the corresponding indicators, and generate a report containing the impact analysis of key parameters for decision-making reference.

[0021] Furthermore, step one specifically includes: Step 1.1, Morphological Prototype Collection and Construction: Collect typical neighborhood building layouts and individual building morphological cases from cold-region cities, extract building layout patterns and building shape types; based on these case morphologies, construct a parametric 3D morphological prototype model library; Step 1.2: Defining and determining the morphological parameters and their ranges; Based on the morphological prototype model, determine the key parameters describing the morphology of the building complex and their physical meaning; the parameters include two parts: one is the parameters of individual buildings within the building complex, including location: (x i ,y i Base length and width (L) i W i The parameters are: building height (H); orientation (O); and building function (F). Secondly, the interrelationships between buildings within the complex include: building density, floor area ratio, building spacing, ventilation corridor width, ventilation corridor connectivity, and street continuity. Statistical calculations are performed on the parameter values ​​for each case, and reasonable value ranges for each parameter are set based on cold-region planning and design requirements and engineering experience. Step 1.3: Establishing a prototype library and parametric representation; Integrate the various morphological parameters defined above into the prototype model extracted in Step 1, and realize the parametric representation of the building complex morphology on the Grasshopper platform; By assigning and changing different parameter combination values, generate different building complex morphological schemes from the prototype library; Finally, establish a morphological prototype library covering multiple layout types and parameter combinations, providing a complete morphological base and parameter support for subsequent automatic scheme generation.

[0022] Furthermore, in step two, hard constraints are set: based on urban planning regulations and the special requirements of cold-region building design, hard constraints that must be strictly met are extracted and quantified into rules that the algorithm can determine. The hard constraints in this invention are building density, floor area ratio, and building spacing. Among these, building spacing includes fire protection requirements and must meet the daylighting spacing standards for cold regions. The hard constraints will be enforced during the optimization process; that is, any solution that does not meet the hard constraints will be considered infeasible and eliminated or penalized.

[0023] The development of soft constraint evaluation indicators: This section, based on the unique climatic conditions of cold regions, is divided into two categories: spatial quality and environmental performance. Spatial quality includes ventilation corridor width, ventilation corridor connectivity, and street continuity. The definitions and calculation formulas for these three indicators are as follows: (1) Width of ventilation corridor w corrw Under the prevailing wind direction d, the minimum effective net width (m) intercepted along the corridor direction within the open zone enclosed by the building's plan projection.

[0024] Let the boundary of the plot be B, and the projection of the foundation plane of the i-th building be... The open area is

[0025] Take a set of intercepts perpendicular to the prevailing wind direction d. },make The width of the ventilation corridor is defined as the minimum value within the range of the through passage: ,in This is a set of cutoff indexes that cover the path of the through channel.

[0026] (2) Connectivity of ventilation corridors C corr Under the prevailing wind direction d, whether the passage from the windward side to the leeward side in the open area O is continuous is characterized by the ratio of the straight-through distance to the shortest feasible passage length (0–1).

[0027] Define the theoretical straight-through distance along the d direction through the plot.

[0028] Within the open region O, find the boundary E on the windward side. in to the leeward boundary Eout Shortest feasible path length ,but

[0029] (3) Street continuity C street : Along the selected street baseline set {r m The direction of the building's street-facing facade is the proportion of the building's continuous street-facing facade length within the evaluation zone to the total street length (0–1).

[0030] For each street baseline Establish an evaluation belt , and all Intersecting building bases projected onto The length of the union of the covered intervals is obtained above. Total length of streets ,but

[0031] Environmental performance includes heating energy consumption intensity during the heating season, effective solar radiation utilization rate in winter, and outdoor wind chill equivalent temperature. These three indicators are defined as follows, and their calculations are obtained through simulation: (4) Heating energy consumption intensity during the heating season: refers to the total amount of heating energy consumed by the neighborhood building complex during the heating season to meet the indoor set temperature, and is the unit area heating energy consumption (kWh / m²) obtained by normalizing the total building area of ​​the building complex. (5) Effective solar utilization rate in winter: The cumulative value of solar radiation energy received per unit area of ​​the public activity area of ​​the neighborhood building complex (kWh / m²).

[0032] (6) Outdoor wind-cold equivalent temperature (J) WCT ): The equivalent perceived temperature (°C) obtained under winter conditions after considering the enhanced heat dissipation effect of wind speed on the human body, is used to characterize the wind chill intensity in cold outdoor spaces.

[0033]

[0034] in, Air temperature; : Pedestrian level wind speed 1.5m.

[0035]

[0036] Where A represents the public activity area; θ represents the wind chill threshold. 20 .

[0037] Furthermore, step three specifically includes: Step 3.1: Automatic scheme generation; using the morphological prototype library established in Step 1, the determined parameter range, and the constructed parametric modeling process, a large number of building group scheme sets are generated in batches through random sampling. Step 3.2, Scheme Indicator Calculation and Simulation: For each automatically generated morphological design scheme, compliance verification is first performed. The hard constraint indicators of the scheme are calculated to see if they meet the set limits. Non-compliant schemes are marked as invalid. Then, for compliant schemes, code is written to calculate the spatial quality indicators. At the same time, based on the Grasshopper platform's Ladybug, Honeybee, and Butterfly, environmental performance indicators are simulated to obtain the indicator results. Step 3.3: Form a training set of scheme data; summarize the parameters and corresponding index calculation results of all candidate schemes to form a complete scheme dataset mapping; this dataset is a training sample set, in which each data point contains the parameter combination of the scheme and the results of various indexes of the scheme.

[0038] Furthermore, step four specifically includes: Step 4.1: Select and build a prediction model; select multiple artificial intelligence algorithms based on the characteristics and scale of the dataset to build a prediction model; the algorithms include MLP, RF, SVR, Catboost, LightGRM, and XGBoost; randomly divide the dataset into training and validation sets, train the selected model to enable it to quickly predict the corresponding index values ​​based on the input design parameters; during the training process, use the grid search method to adjust the model hyperparameters to improve prediction accuracy, and evaluate the error level of the model through the validation set to ensure that each prediction model has sufficient accuracy; Step 4.2, Multi-model Comparison and Validation: After the prediction model training converges and achieves the expected accuracy on the validation set, conduct a multi-algorithm comparison; using R... 2 The accuracy of the results is verified using three metrics: RMSE, MAE, and RMSE, and the optimal model is selected accordingly.

[0039] Furthermore, step five specifically includes: Step 5.1: Multi-objective optimization initialization; Select the NSGA-Ⅲ algorithm and set the parameters, including population size, number of iterations, crossover and mutation probability, etc. Use the scheme set as the initial population, and encode each individual as a set of building cluster morphology parameters. The corresponding indicators are obtained by the trained prediction model through rapid prediction. Step 5.2, Evolutionary Iteration Process: In each generation of optimization iteration, individuals in the population are selected, crossovered, and mutated based on multi-objective fitness to generate new candidate solutions. For each newly generated solution individual, its various index values ​​are first quickly evaluated through a prediction model, and then its compliance with hard constraints is checked: if an individual violates the hard constraints, the iteration is restarted; compliant individuals participate in Pareto sorting and crowding calculation operations according to the requirements of the optimization algorithm to determine their survival probability in the population. The non-dominated sorting mechanism is used to retain individuals that perform well in multi-objectives and eliminate inferior individuals. Step 5.3, Convergence Judgment and Termination: Repeat the evolutionary iteration process and continuously update the population; the iteration ends when the preset termination condition is reached; the termination condition is set according to the optimization requirements, such as reaching a predetermined upper limit of generations or the Pareto front of the population tending to stagnate within a few generations. After the iteration ends, all non-dominated solutions in the final population are extracted as the result scheme set of this optimization; each scheme in the obtained Pareto front scheme set has satisfied the hard constraints and achieved different trade-off combinations between spatial quality and environmental performance indicators.

[0040] Furthermore, step six specifically includes: We select interpretable methods and obtain the importance of global features; for the Pareto solution set obtained by optimization, we use interpretable artificial intelligence technology to perform attribution analysis on the solution performance; we use SHAP value analysis method to calculate the contribution value of each morphological parameter to each target index, and use PDP diagram to calculate the threshold and nonlinear response of morphological parameters to each target value. Through these mechanisms, we use interpretive methods to help designers understand the causes of optimization solutions.

[0041] Furthermore, step seven specifically includes: Step 7.1: Output the optimized scheme set and indicators; Output the final Pareto front scheme set and list the target indicator values ​​for each scheme; Make it easy for designers to view and retrieve the schemes through lists, tables or 3D model scheme integration files; For each scheme, clearly give its key parameter configuration and the achieved spatial quality index value and environmental performance index value. Step 7.2: Generate an interpretable optimization report; Based on the analysis results, an interpretable report document containing key information about the optimization process is automatically generated; This report, combined with interpretable methods, displays the main parameters and their importance ranking, and summarizes the direction of the influence of design parameters on each indicator.

[0042] Example This invention proposes a method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence. The method includes the following steps: Step 1: Establish a morphological prototype library and parameter system. Construct a prototype model library of the 3D morphology of the cold-region neighborhood building complex, establish a morphological parameterization representation system, and define the key morphological parameters and attributes of each building in the complex: Location: (x i ,y i ), Base length and width (L) i W i The building's dimensions are: height (H), orientation (O), and function (F). The range of these values ​​is extracted to provide a basis for subsequent design generation.

[0043] Step two involves determining the constraints and evaluation index system for the design scheme. This part is divided into two categories: hard constraints and soft performance objectives. Hard constraints are the urban regulations that the design site must meet, which in this invention are building density, floor area ratio, or building spacing. Soft performance objectives, in the context of cold-region design, are divided into two categories: spatial quality and environmental performance. Spatial quality includes ventilation corridor width, ventilation corridor connectivity, and street continuity. Environmental performance includes heating energy consumption intensity during the heating season, effective solar radiation utilization rate in winter, and outdoor wind chill equivalent temperature.

[0044] Step 3: Generate schemes and form an indicator dataset. Based on the parameter system in Step 1, a morphological scheme dataset is generated through random sampling, and batch generation is performed using parametric modeling methods. Indicators are calculated or simulated for each scheme to obtain the constraints and evaluation indicator system results, forming a mapping between scheme parameter combinations and the indicator system.

[0045] Step four: Establish a predictive model. Based on the result mapping formed in steps one through three, construct a data-driven artificial intelligence predictive model to achieve rapid prediction of the target value for each scheme, replacing high-cost simulation calculations and accelerating subsequent morphological optimization iterations.

[0046] Step 5, Multi-objective Pareto Evolutionary Optimization. Using the initial set of solutions as the initial population, an iterative search is performed using the Multiple Non-Dominated Sorting Genetic Algorithm (NSGA-III). In this process, the solutions must first meet the compliance requirements of the hard constraints. On this basis, a multi-objective trade-off is made between the two soft performance objectives of space quality and environmental performance, outputting a set of non-dominated Pareto front solutions.

[0047] Step six: For the optimal three-dimensional form scheme of the building complex obtained through multi-objective optimization, the present invention further introduces an interpretable artificial intelligence (XAI) module. Using model interpretation technology based on SHAP value, LIME and PDP graph, the contribution of design parameters to performance indicators and their mechanism are quantified, and the graph or visualization results of the influence of key parameters are output.

[0048] Step 7: Output the Pareto scheme set and its corresponding index results from Step 5; combine with Step 6 to generate an interpretable report to support design decisions and planning reviews.

[0049] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence.

[0050] 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 method for optimizing the three-dimensional morphology of cold-region neighborhood building complexes based on interpretable artificial intelligence.

[0051] The memory in this application embodiment can be volatile memory or non-volatile memory, or it can include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0052] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as 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, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0053] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0054] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The 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 gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as execution by a hardware decoding processor, or as a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0055] The above provides a detailed description of the three-dimensional morphology optimization method for cold-region neighborhood building complexes based on interpretable artificial intelligence proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An explainable artificial intelligence-based three-dimensional form optimization method for cold neighborhood building clusters, characterized in that, The method includes: Step 1: Establish a morphological prototype library and parameter system; based on the design requirements of urban block-scale building complexes in cold regions, pre-construct a three-dimensional morphological prototype model library, establish a corresponding morphological parameterization representation system, and statistically analyze parameter ranges to describe the morphological characteristics of the building complex. Step two: Determine the constraints and evaluation index system; transform the mandatory requirements of urban planning into hard constraints, and at the same time construct a soft performance index system that reflects the spatial quality and environmental performance of cold-region neighborhood building clusters, which together serve as subsequent optimization objectives. Step 3: Generate a batch of scheme combinations and construct an indicator data dataset; construct a large number of morphological schemes based on random sampling, and output the evaluation index results of each scheme through calculation and simulation, and summarize them to form a training dataset; Step 4: Build a prediction model; use the above dataset to train the prediction model to achieve rapid evaluation and prediction of the scheme's indicators, thereby accelerating the subsequent optimization process. Step 5: Multi-objective Pareto evolution optimization; The NSGA-III algorithm is used to iteratively optimize and search for multiple objectives under the premise of satisfying hard constraints, so as to obtain a set of Pareto optimal solutions. Step 6: Interpretable AI Attribution Analysis; Perform interpretable analysis on the obtained optimal solution set to identify the dominant design parameters and their impact on various objectives; Step 7: Output the optimization results and generate an explanatory report; finally, output the Pareto solution set obtained from the optimization and the corresponding indicators, and generate a report containing the impact analysis of key parameters for decision-making reference.

2. The method of claim 1, wherein, Step one specifically includes: Step 1.1, Morphological Prototype Collection and Construction: Collect typical neighborhood building layouts and individual building morphological cases from cold-region cities, extract building layout patterns and building shape types; based on these case morphologies, construct a parametric 3D morphological prototype model library; Step 1.2: Defining and determining the morphological parameters and their ranges; Based on the morphological prototype model, determine the key parameters describing the morphology of the building complex and their physical meaning; The parameters include two parts: one is the parameters of individual buildings within the building complex; the other is the interrelationship between buildings within the building complex; Calculate and statistically analyze the parameter values ​​for each case, and combine the planning and design requirements of cold regions with engineering experience to set a reasonable range for each parameter. Step 1.3: Establishing a prototype library and parametric representation; Integrate the various morphological parameters defined above into the prototype model extracted in Step 1, and realize the parametric representation of the building complex morphology on the Grasshopper platform; By assigning and changing different parameter combination values, generate different building complex morphological schemes from the prototype library; Finally, establish a morphological prototype library covering multiple layout types and parameter combinations, providing a complete morphological base and parameter support for subsequent automatic scheme generation.

3. The method according to claim 1, characterized in that, In step two, the hard constraints are building density, floor area ratio, or building spacing; the soft performance index system is divided into two categories in the context of cold region design: spatial quality and environmental performance. Spatial quality includes ventilation corridor width, ventilation corridor connectivity, and street continuity; environmental performance includes heating energy consumption intensity during the heating season, effective solar utilization rate in winter, and outdoor wind chill equivalent temperature.

4. The method according to claim 1, characterized in that, Step three specifically includes: Step 3.1: Automatic scheme generation; using the morphological prototype library established in Step 1, the determined parameter range, and the constructed parametric modeling process, a large number of building group scheme sets are generated in batches through random sampling. Step 3.2, Scheme Indicator Calculation and Simulation: For each automatically generated morphological design scheme, compliance verification is first performed. The hard constraint indicators of the scheme are calculated to see if they meet the set limits. Non-compliant schemes are marked as invalid. Then, for compliant schemes, code is written to calculate the spatial quality indicators. At the same time, based on the Grasshopper platform's Ladybug, Honeybee, and Butterfly, environmental performance indicators are simulated to obtain the indicator results. Step 3.3: Form a training set of scheme data; summarize the parameters and corresponding index calculation results of all candidate schemes to form a complete scheme dataset mapping; this dataset is a training sample set, in which each data point contains the parameter combination of the scheme and the results of various indexes of the scheme.

5. The method according to claim 1, characterized in that, Step four specifically includes: Step 4.1: Select and build a prediction model; select multiple artificial intelligence algorithms based on the characteristics and scale of the dataset to build a prediction model; the algorithms include MLP, RF, SVR, Catboost, LightGRM, and XGBoost; randomly divide the dataset into training and validation sets, train the selected model to enable it to quickly predict the corresponding index values ​​based on the input design parameters; during the training process, use the grid search method to adjust the model hyperparameters to improve prediction accuracy, and evaluate the error level of the model through the validation set to ensure that each prediction model has sufficient accuracy; Step 4.2, multi-model comparison and verification; when the prediction model training converges and reaches the expected accuracy on the validation set, carry out multi-algorithm comparison; use R 2 , RMSE, MAE three indexes to verify the accuracy of the results, and select the optimal model.

6. The method according to claim 1, characterized in that, Step five specifically includes: Step 5.1: Multi-objective optimization initialization; Select the NSGA-Ⅲ algorithm and set the parameters. Use the scheme set as the initial population. Each individual is encoded as a set of building group morphology parameters, and the corresponding index is obtained by the trained prediction model for rapid prediction. Step 5.2, Evolutionary Iteration Process: In each generation of optimization iteration, individuals in the population are selected, crossovered, and mutated based on multi-objective fitness to generate new candidate solutions. For each newly generated solution individual, its various index values ​​are first quickly evaluated through a prediction model, and then its compliance with hard constraints is checked: if an individual violates the hard constraints, the iteration is restarted; compliant individuals participate in Pareto sorting and crowding calculation operations according to the requirements of the optimization algorithm to determine their survival probability in the population. The non-dominated sorting mechanism is used to retain individuals that perform well in multi-objectives and eliminate inferior individuals. Step 5.3, Convergence Judgment and Termination: Repeat the evolutionary iteration process and continuously update the population; the iteration ends when the preset termination condition is reached; the termination condition is set according to the optimization requirements. After the iteration ends, all non-dominated solutions in the final population are extracted as the result scheme set of this optimization; each scheme in the obtained Pareto front scheme set has satisfied the hard constraints and achieved different trade-off combinations between spatial quality and environmental performance indicators.

7. The method according to claim 1, characterized in that, Step six specifically includes: We select interpretable methods and obtain the importance of global features; for the Pareto solution set obtained by optimization, we use interpretable artificial intelligence technology to perform attribution analysis on the solution performance; we use SHAP value analysis method to calculate the contribution value of each morphological parameter to each target index, and use PDP diagram to calculate the threshold and nonlinear response of morphological parameters to each target value. Through these mechanisms, we use interpretive methods to help designers understand the causes of optimization solutions.

8. The method according to claim 1, characterized in that, Step seven specifically includes: Step 7.1: Output the optimized scheme set and indicators; Output the final Pareto front scheme set and list the target indicator values ​​for each scheme; Make it easy for designers to view and retrieve the schemes through lists, tables or 3D model scheme integration files; For each scheme, clearly give its key parameter configuration and the achieved spatial quality index value and environmental performance index value. Step 7.2: Generate an interpretable optimization report; Based on the analysis results, an interpretable report document containing key information about the optimization process is automatically generated; This report, combined with interpretable methods, displays the main parameters and their importance ranking, and summarizes the direction of the influence of design parameters on each indicator.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-8.

Images