Building environment simulation and optimization method

Abstract

The invention provides a building environment simulation and optimization method, and belongs to the technical field of building simulation, and the method comprises the steps: combining building rational properties associated with user behavior demands, combining building simulation dimensions set by environment data, and fusing the building simulation dimensions into a building three-dimensional geometric model to obtain a first building model, carrying out single-dimensional simulation and full-dimensional simulation, constructing an individual error simulation graph corresponding to the building simulation dimension and a comprehensive error simulation graph, and carrying out synchronous evolution and abnormal intervention on the comprehensive error simulation graph and the individual error simulation graph under each building simulation dimension by using a minimum time sequence unit; and a first optimization scheme is generated for optimization and practical application, and secondary optimization is continued according to the difference between reality and simulation. And the simulation efficiency is improved, so that the building environment can well meet user requirements.

Description

Technical Field

[0001] This invention relates to the field of building simulation technology, and in particular to a method for simulating and optimizing building environments. Background Technology

[0002] Building simulation refers to the use of computer technology to construct digital models of buildings. By inputting environmental parameters (such as climate and external loads), building physical properties (such as materials and structural forms), and user behavior characteristics (such as usage periods and functional requirements), it simulates various performance aspects of a building throughout its entire lifecycle (design, construction, and operation and maintenance), including energy consumption, lighting, ventilation, temperature and humidity, and acoustics. Its core is to replace physical testing with digital simulation, enabling the prediction, analysis, and optimization of dynamic changes in the building environment, providing data support for building design and operation and maintenance decisions. Traditional building environment simulation has three major limitations: First, simulation models are mostly statically constructed based on building geometric features, ignoring the dynamic correlation between user behavior and building physical performance, leading to a disconnect between simulation results and actual usage scenarios. Second, multi-dimensional simulation data lacks effective alignment and fusion mechanisms, and the correlation analysis between single-dimensional errors and comprehensive errors is insufficient, making it difficult to locate cross-dimensional coupling problems. Third, the optimization process is mostly a single static adjustment, unable to adapt to the dynamic changes in the environment and user behavior throughout the building's lifecycle, ultimately resulting in low simulation efficiency and a building environment that fails to adequately meet user needs.

[0003] Therefore, this invention proposes a method for simulating and optimizing the building environment. Summary of the Invention

[0004] This invention provides a method for simulating and optimizing the building environment to solve the aforementioned technical problems.

[0005] This invention provides a method for simulating and optimizing a built environment, comprising:

[0006] Step 1: Collect architectural design drawings related to the built environment to construct a three-dimensional geometric model of the building. Combine the building's physical performance related to user behavior needs with the building simulation dimensions set by environmental data and integrate them into the three-dimensional geometric model of the building to obtain the first building model.

[0007] Step 2: Perform single-dimensional and full-dimensional simulations on the first building model, and capture the first process data of each single-dimensional simulation and the global process data of the full-dimensional simulation.

[0008] Step 3: Perform a first alignment process between the first process data and the setting standard of the single-dimensional simulation to construct a separate error simulation diagram for the corresponding building simulation dimension. At the same time, perform a second alignment process between the global process data and all first process data to construct a comprehensive error simulation diagram.

[0009] Step 4: Synchronously evolve and intervene in anomalies of the comprehensive error simulation map and the individual error simulation map under each building simulation dimension in the smallest time sequence unit to generate the first optimization scheme, and perform initial optimization on the first building model and its actual application after initial optimization. Then, based on the difference between the actual running data and the simulation results after initial optimization, continue to optimize the model after initial optimization.

[0010] Preferably, constructing a three-dimensional geometric model of the building includes:

[0011] Each structural component in the architectural design drawings is automatically extracted to generate a component data block with dimension annotations;

[0012] The parametric components that match the component data blocks are matched from the component attribute database, and the corresponding parametric components are assembled using the design framework of the architectural design drawings to obtain a three-dimensional geometric model of the building.

[0013] Preferably, the building's physical performance is determined in relation to user behavior needs, and building simulation dimensions are set in conjunction with environmental data, including:

[0014] The user behavior requirements are analyzed to obtain a first state vector based on static behavior and a second state vector based on dynamic behavior. The first state vector and the second state vector are then extracted and concatenated for the same sub-requirement scenario to obtain a concatenated sub-matrix. The satisfaction coefficient of the corresponding sub-requirement scenario is determined. At the same time, the static behavior data and dynamic behavior data of the user in the building space are collected and scenario analysis is performed to obtain the dynamic features associated with each static feature. A reference association array for the corresponding static features is constructed, wherein the reference association array contains the association direction between the static features and each dynamic feature.

[0015] Based on the theoretical mapping relationship between sub-demand scenarios and each static feature, the matching reference association array is retrieved, and the spliced ​​sub-matrix is ​​optimized by combining the satisfaction coefficient to obtain a new sub-matrix;

[0016] Taking each preset building physics index as the entry point, interference analysis is performed on each new submatrix in turn to determine the index that affects the corresponding new submatrix, which is regarded as the first index, and the building physics performance of the corresponding new submatrix is ​​obtained based on the index-performance comparison table.

[0017] The building physics performance corresponding to the first metric that is most affected by each new submatrix is ​​retained and regarded as the first performance.

[0018] Meanwhile, the building physical performance corresponding to the remaining first index of each new sub-matrix is ​​intersected to obtain the second performance, wherein the second performance is the result of the building physical performance under the corresponding remaining first index appearing twice or more;

[0019] Based on the first and second performance metrics, and combined with environmental data, the building simulation dimensions are set.

[0020] Preferably, based on the first and second performance metrics, and in conjunction with environmental data, the building simulation dimensions are set, including:

[0021] The environmental data is decomposed based on the attribute characteristics of each performance component to construct spatial entities;

[0022] After encoding and converting each spatial entity, it is mapped to an entity lookup table, and each spatial entity has a mapping occupancy coefficient in the entity lookup table based on the lookup area under each preset simulation dimension;

[0023]

[0024] Among them, R i represents the mapping occupancy coefficient of the reference region in the i-th preset simulation dimension; mi represents the number of spatial entities involved in the reference region in the i-th preset simulation dimension; M j Q represents the number of occupied units of the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j δ represents the weight of the performance corresponding to the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j represents the auxiliary function of the j-th spatial entity; ai represents the entity coefficient of the ith preset simulation dimension; Ni is the total number of elements in the reference region of the ith preset simulation dimension;

[0025] If the mapping occupancy coefficient is greater than the set coefficient of the corresponding reference area, then the corresponding preset simulation dimension will be regarded as the building simulation dimension.

[0026] Otherwise, the corresponding preset simulation dimension will be removed.

[0027] Preferably, the matching reference association array is retrieved, and the concatenated submatrix is ​​optimized using the satisfaction coefficients to obtain a new submatrix, including:

[0028] Based on the theoretical mapping relationship between sub-demand scenarios and static features, a correlation strength tensor matrix is ​​constructed, where the matrix elements quantify the correlation strength between sub-demand scenarios and static features in numerical form.

[0029] The reference association arrays are prioritized based on the numerical values ​​of each element in the association strength tensor matrix, and reference association arrays with association strength exceeding a preset threshold are selected as the core optimization arrays.

[0030] The spliced ​​submatrix is ​​dimension-matched with the core optimization array. Mismatched dynamic feature dimensions are normalized to a spatial scale that is homologous with the static feature dimensions, generating a dimension-aligned intermediate matrix.

[0031] The elements in the dimension-aligned intermediate matrix are corrected based on the comprehensive correlation direction in the core optimization array. Specifically, if the comprehensive correlation direction is positive, then... Enhance the association weights between corresponding static and dynamic features, where G(fs) is a weight incremental function based on the positive association direction fs; mx is the satisfaction coefficient; and yx is a preset coefficient.

[0032] If the overall correlation direction is negative, then according to Weaken the corresponding association weights, where G(ff) is a weight decay function based on the negative direction of the comprehensive association ff;

[0033] Calculate the goodness of fit between the corrected dimension-aligned intermediate matrix and the theoretical mapping relationship. If the goodness of fit does not meet the preset standard, retrieve the secondary priority supplementary array from the reference associated array for secondary correction until the goodness of fit meets the requirements and obtain the new submatrix.

[0034] Preferably, a separate error simulation diagram corresponding to the building simulation dimension is constructed, including:

[0035] Based on the first alignment processing result, the trend term and fluctuation term in the time dimension and the regional distribution features in the spatial dimension are extracted, a feature dual matrix is ​​constructed, and the relative deviation, trend error and fluctuation error between the first process data and the set standard are calculated to form an error vector.

[0036] The error vector is mapped to the corresponding spatial region of the three-dimensional geometric model, and the error type is distinguished based on a preset legend to generate a corresponding individual error simulation diagram.

[0037] Preferably, a comprehensive error simulation diagram is constructed, including:

[0038] The global process data is sequentially aligned and compared with the setting standards of all single-dimensional simulations to generate the first global error simulation diagram.

[0039] The global process data is compared with the first process data under all building simulation dimensions to generate the first global deviation simulation diagram.

[0040] The first global deviation simulation map and the first global error simulation map are projected onto the three-dimensional geometric model of the building, respectively, to determine the building error form based on each building simulation dimension. When the building error form has an interactive form in other dimensions, auxiliary labels are set for the building error form based on the interactive form.

[0041] Based on the architectural error patterns and auxiliary labels under the corresponding architectural simulation dimensions, the corresponding dimension layers in the first global error simulation map are adjusted to obtain the comprehensive error simulation map.

[0042] Preferably, the comprehensive error simulation map and the individual error simulation map under each building simulation dimension are synchronously evolved and anomaly intervened in using the smallest time-series unit to generate a first optimization scheme, including:

[0043] Configure a unified timing reference clock for the integrated error simulation graph and each individual error simulation graph, extract the error feature vector of each smallest timing unit, and construct the error feature timing stream;

[0044] The internal dynamic correlation analysis and inter-time dynamic correlation analysis are performed on the time-series flow of the error characteristics to calculate the real-time correlation matrix between the comprehensive error and each individual error.

[0045] The error change rate prediction model, trained based on historical evolution data, generates a dynamic warning slope for each smallest time series unit.

[0046] When a dimensional error triggers a static threshold in a single error simulation graph, if the overall error does not exceed the dynamic warning slope, the corresponding dimension's error correction rule base will be automatically retrieved to generate and record the in-dimensional optimization instructions.

[0047] If the overall error exceeds the dynamic warning slope, then, based on the strong correlation errors existing in the real-time correlation matrix and the generation trajectory of the strong correlation errors, the parameter combination that minimizes the error collaboration is solved and recorded.

[0048] If a dimensional error exists in a single error simulation graph but does not trigger the static threshold, it is determined that there is no error in the corresponding dimension and recorded.

[0049] The first optimization scheme is generated based on the recorded results.

[0050] Compared with the prior art, the beneficial effects of this application are as follows:

[0051] By integrating user behavior and environmental data to construct a dynamic simulation model, and combining multi-dimensional error correlation analysis and time-series anomaly intervention, the accuracy and optimization efficiency of building environment simulation are significantly improved, enabling dynamic adaptation and efficient operation and maintenance throughout the entire building lifecycle, thus meeting user needs and experience.

[0052] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0053] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0055] Figure 1 This is a flowchart of a building environment simulation and optimization method according to an embodiment of the present invention. Detailed Implementation

[0056] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0057] This invention provides a method for simulating and optimizing a building environment, such as... Figure 1 As shown, it includes:

[0058] Step 1: Collect architectural design drawings related to the built environment to construct a three-dimensional geometric model of the building. Combine the building's physical performance related to user behavior needs with the building simulation dimensions set by environmental data and integrate them into the three-dimensional geometric model of the building to obtain the first building model.

[0059] Step 2: Perform single-dimensional and full-dimensional simulations on the first building model, and capture the first process data of each single-dimensional simulation and the global process data of the full-dimensional simulation.

[0060] Step 3: Perform a first alignment process between the first process data and the setting standard of the single-dimensional simulation to construct a separate error simulation diagram for the corresponding building simulation dimension. At the same time, perform a second alignment process between the global process data and all first process data to construct a comprehensive error simulation diagram.

[0061] Step 4: Synchronously evolve and intervene in anomalies of the comprehensive error simulation map and the individual error simulation map under each building simulation dimension in the smallest time sequence unit to generate the first optimization scheme, and perform initial optimization on the first building model and its actual application after initial optimization. Then, based on the difference between the actual running data and the simulation results after initial optimization, continue to optimize the model after initial optimization.

[0062] In this embodiment, architectural design drawings refer to architectural plans, construction drawings, structural component details, etc., including the dimensions, materials, and location information of structures such as walls, beams, columns, doors, and windows (e.g., CAD floor plan of an office building, Revit component list).

[0063] A building's 3D geometric model is a 3D digital model built based on design drawings. It includes the spatial location, dimensions, and topological relationships of components (e.g., using Revit software to parameterize components such as walls and floors in CAD drawings to generate a BIM 3D model).

[0064] User behavior needs refer to the activity characteristics and needs of users in architectural spaces, which are divided into static behaviors (long-term activities in fixed locations) and dynamic behaviors (movement or temporary activities within the space) (e.g., "8-hour fixed office work" for office users is a static behavior, while "temporary discussion in the meeting room" is a dynamic behavior).

[0065] Building physical performance refers to the physical characteristics of a building that meet the needs of user behavior and is directly related to user experience (e.g., office areas require properties such as "uniform lighting" and "summer room temperature of 24-26℃", corresponding to the comfort needs of static behavior; corridors require "efficient ventilation", corresponding to the mobility needs of dynamic behavior).

[0066] Environmental data refers to the natural environmental parameters of the building's exterior and surroundings (e.g., outdoor temperature and humidity, wind speed, solar radiation intensity, noise level, etc., such as an average daily temperature of 32℃ and wind speed of 1.5m / s in a city during summer).

[0067] The first building model is a dynamic model that integrates the three-dimensional geometric model of the building, the physical performance of the building, and the simulation dimensions of the building (e.g., embedding simulation parameters of dimensions such as "energy consumption" and "comfort" into the BIM model, and associating boundary conditions with the user's office hours (9:00-18:00).

[0068] Single-dimensional simulation is a simulation analysis performed independently on a single building simulation dimension (e.g., simulating only the "energy consumption dimension" and calculating the energy consumption changes of the air conditioning system at different times).

[0069] Full-dimensional simulation is a collaborative simulation that simultaneously performs simulations of all dimensions of building simulation, reflecting the mutual influence between dimensions (e.g., simultaneously simulating the dimensions of "energy consumption", "lighting" and "ventilation", and analyzing the coupling relationship between air conditioning energy consumption and window ventilation).

[0070] The first process data is dynamic data generated in the single-dimensional simulation, reflecting the real-time changes in that dimension (e.g., "hourly air conditioning power consumption" in the energy consumption dimension (12kWh at 8:00 and 18kWh at 12:00)).

[0071] Global process data is comprehensive data generated in full-dimensional simulation, reflecting the overall state of multi-dimensional interactions (e.g., "At 12:00, energy consumption is 18kWh, room temperature is 26℃, and window ventilation is 0.5m³"). 3 (The associated data of " / h").

[0072] The first alignment process involves comparing the first-process data of a single dimension with the industry standards or design goals of that dimension to unify the data format and time scale.

[0073] Individual error simulation plots are visualizations that intuitively show the spatial distribution and temporal changes of single-dimensional errors.

[0074] The second alignment process involves comparing the global process data with all single-dimensional first process data to analyze the correlation of multi-dimensional errors.

[0075] The comprehensive error simulation diagram is a visualization that shows the multi-dimensional error coupling relationship, including single-dimensional errors and cross-dimensional effects.

[0076] The minimum time unit is the minimum time interval between simulation and optimization, which is set according to the dynamic characteristics of the building (e.g., office buildings are set in 15-minute intervals to capture short-term fluctuations in energy consumption, temperature and humidity in real time).

[0077] Synchronous evolution means that the comprehensive error simulation map and the individual error simulation map are dynamically updated with the smallest time unit to reflect the real-time changes in error (e.g., the energy consumption error map and the comprehensive error map are updated every 15 minutes to observe whether excessive energy consumption leads to an increase in comfort error).

[0078] Abnormal intervention is to automatically trigger adjustment strategies when the error exceeds the threshold (e.g., when the energy consumption error exceeds 20%, the air conditioner setting temperature is automatically reduced by 1℃; when the comprehensive error display shows an abnormal coupling of "energy consumption-ventilation", the air conditioner power and window opening angle are adjusted simultaneously).

[0079] The first optimization plan is a preliminary adjustment strategy generated based on the results of abnormal intervention (e.g., "Set the air conditioning temperature in the office area to 25°C from 9:00 to 12:00 and turn off unnecessary lighting on the west side to reduce energy consumption").

[0080] The initial optimization involves applying the first optimization scheme to the first building model and updating the model parameters (e.g., modifying the air conditioning operation schedule in EnergyPlus and resimulating).

[0081] Actual operating data refers to monitoring data collected during actual use of the building (e.g., "office area room temperature 25.5℃, energy consumption 15kWh at 10:00" collected by temperature and humidity sensors and smart meters).

[0082] Secondary optimization involves comparing actual operating data with the simulation results after the initial optimization and correcting the model parameters (e.g., if the actual energy consumption is 10% higher than the simulation value, the efficiency coefficient of the air conditioning equipment is adjusted, and a new optimization scheme is generated).

[0083] The beneficial effects of the above technical solution are: by integrating user behavior and environmental data to construct a dynamic simulation model, and combining multi-dimensional error correlation analysis and time-series anomaly intervention, the accuracy and optimization efficiency of building environment simulation are significantly improved, realizing dynamic adaptation and efficient operation and maintenance throughout the entire building life cycle, and meeting the needs and experience of users.

[0084] This invention provides a method for building environment simulation and optimization, which constructs a three-dimensional geometric model of a building, including:

[0085] Each structural component in the architectural design drawings is automatically extracted to generate a component data block with dimension annotations;

[0086] The parametric components that match the component data blocks are matched from the component attribute database, and the corresponding parametric components are assembled using the design framework of the architectural design drawings to obtain a three-dimensional geometric model of the building.

[0087] In this embodiment, structural components are the basic units that constitute the main body of the building. They have clear geometric shapes and functional attributes, including walls, beams, columns, floor slabs, doors and windows, stairs, etc. (e.g., 240mm thick aerated concrete walls, reinforced concrete beams with a cross-sectional dimension of 300mm×500mm).

[0088] Automatic extraction is the process of automatically identifying and separating each structural component from architectural design drawings using computer vision and image recognition technology.

[0089] Dimensioning is the annotation information used in drawings to represent the geometric parameters of structural components, including length, width, height, thickness, etc. (For example, "240mm" for a wall indicates its thickness, and "300×500" for a beam indicates its width × height).

[0090] A component data block is a structured data unit formed by integrating automatically extracted structural component information (including type, size, and location coordinates). It is usually stored in JSON or XML format. For example, a wall component data block is {Type: Exterior Wall, Thickness: 240mm, Material: Aerated Concrete, Coordinate Range: {x1:0,y1:0,x2:10000,y2:240}}, where the coordinate unit is mm.

[0091] The component attribute database is a database that stores parametric information of various structural components, including the standard dimensions, material properties, mechanical parameters, and 3D model templates of the components.

[0092] In this embodiment, the parametric component is a standardized three-dimensional component model with adjustable parameters. By modifying the parameters (such as length and thickness), it can be adapted to different design requirements without the need for remodeling.

[0093] The design framework is the spatial layout logic and component connection relationship implicit in architectural design drawings, including the relative position of components, assembly sequence, and topological structure. Assembly is the process of splicing matching parametric components into a whole three-dimensional model according to the spatial logic of the design framework and their positional relationships.

[0094] A three-dimensional geometric model of a building is a three-dimensional digital model formed through assembly, which fully presents the spatial structure and component details of the building.

[0095] The beneficial effects of the above technical solution are: by automatically extracting structural components and matching them with parametric components, the generated three-dimensional geometric model combines standardization and flexibility, providing an accurate digital foundation for subsequent building simulation and performance analysis.

[0096] This invention provides a method for building environment simulation and optimization, which determines the physical performance of buildings in relation to user behavior needs and sets building simulation dimensions based on environmental data, including:

[0097] The user behavior requirements are analyzed to obtain a first state vector based on static behavior and a second state vector based on dynamic behavior. The first state vector and the second state vector are then extracted and concatenated for the same sub-requirement scenario to obtain a concatenated sub-matrix. The satisfaction coefficient of the corresponding sub-requirement scenario is determined. At the same time, the static behavior data and dynamic behavior data of the user in the building space are collected and scenario analysis is performed to obtain the dynamic features associated with each static feature. A reference association array for the corresponding static features is constructed, wherein the reference association array contains the association direction between the static features and each dynamic feature.

[0098] Based on the theoretical mapping relationship between sub-demand scenarios and each static feature, the matching reference association array is retrieved, and the spliced ​​sub-matrix is ​​optimized by combining the satisfaction coefficient to obtain a new sub-matrix;

[0099] Taking each preset building physics index as the entry point, interference analysis is performed on each new submatrix in turn to determine the index that affects the corresponding new submatrix, which is regarded as the first index, and the building physics performance of the corresponding new submatrix is ​​obtained based on the index-performance comparison table.

[0100] The building physics performance corresponding to the first metric that is most affected by each new submatrix is ​​retained and regarded as the first performance.

[0101] Meanwhile, the building physical performance corresponding to the remaining first index of each new sub-matrix is ​​intersected to obtain the second performance, wherein the second performance is the result of the building physical performance under the corresponding remaining first index appearing twice or more;

[0102] Based on the first and second performance metrics, and combined with environmental data, the building simulation dimensions are set.

[0103] In this embodiment, static behavior is a long-term, stable activity performed by a user in a fixed spatial location with a small range of spatial movement (e.g., working at a workstation, sitting for a long time in a meeting room, or resting on a sofa). The first state vector is a numerical vector representing the key features of static behavior. The vector dimensions include the duration of the behavior, the frequency of occurrence, and the spatial location (e.g., the first state vector for an office scenario is [8 hours / day (duration), 1 time / day (frequency), workstation area (location)]).

[0104] Dynamic behavior refers to the mobile and temporary activities that users engage in within a building space, where spatial location changes over time (e.g., walking in a corridor, entering or exiting an elevator, going to the tea room to get water, or switching between different meeting rooms). The second state vector is a numerical vector representing the key features of dynamic behavior. Vector dimensions include movement distance, activity duration, and time period (e.g., the second state vector for an office scenario is [500 meters / day (movement distance), 0.5 hours / day (total duration), 10:00-16:00 (time period)]).

[0105] A sub-demand scenario is a collection of user behaviors with the same functional goal (e.g., "office scenario" includes sub-behaviors such as "working at a workstation" and "discussion in a meeting room", all of which serve the need to "complete work tasks").

[0106] The satisfaction coefficient measures the degree of matching between the behavioral data in the spliced ​​submatrix and the theoretical goal of the sub-demand scenario. (For example, the theoretical requirement for an office scenario is "static duration ≥ 6 hours", but the actual duration is 8 hours, so the satisfaction coefficient = 8 / 6 ≈ 1.3; if the actual duration is 4 hours, the satisfaction coefficient = 4 / 6 ≈ 0.67).

[0107] Static behavior data is a quantitative record describing static behavior (e.g., the daily duration of working at a workstation, the number of times spent in the rest area).

[0108] Dynamic behavior data is a quantitative record that describes dynamic behavior (e.g., the number of steps taken in a corridor or the frequency of elevator use).

[0109] Scene analysis involves clustering and correlation analysis of static and dynamic behavioral data to identify the correspondence between behaviors and scenes (for example, analysis may find that "high static time from 9:00 to 12:00 + low movement distance" corresponds to "focused office scene").

[0110] Static characteristics are the core attributes of static behavior (e.g., duration, spatial fixity, and behavioral intensity, such as "the continuous stability of sitting posture while working").

[0111] Dynamic characteristics are the core attributes of dynamic behavior (e.g., movement speed, spatial span, behavior frequency, such as "average interval time between trips to and from the tea room").

[0112] The reference association array is an array that records the association relationship between static features and dynamic features. It includes the association direction (positive / negative) and association strength (for example, the association array between static office duration (static feature) and number of trips to the break room (dynamic feature) is [positive association, 0.8 (strength)], indicating that the longer the office work, the more frequently people go to the break room).

[0113] The theoretical mapping relationship between sub-demand scenarios and static features is the pre-defined correspondence rule between sub-demand scenarios and static features (for example, "office scenario" theoretically requires "static duration ≥ 6 hours" and "spatial fixity ≥ 0.9", where 0.9 means 90% of the time is spent in a fixed area).

[0114] The new submatrix is ​​an optimized matrix obtained by correcting the deviations in the spliced ​​submatrix by referring to the associative array and adjusting the weights by combining the satisfaction coefficient (for example, if the satisfaction coefficient of the office scenario is 0.67 (insufficient static time), and the reference associative array shows "static time is negatively correlated with moving distance", then the weight of moving distance is reduced to generate a new submatrix that better fits the theoretical mapping).

[0115] Preset building physics indicators are parameters that measure the physical characteristics of the building environment, including temperature, humidity, light intensity, noise level, and air velocity.

[0116] Interference analysis examines the degree of influence of pre-defined building physics indicators on the new sub-matrix (sub-demand scenario). (For example, analysis of the new sub-matrix for the office scenario found that temperature fluctuations have an impact of 0.7 (the highest) on static office time, while light intensity has an impact of 0.5).

[0117] The primary indicator is the building physics indicator that has the greatest impact on the new submatrix (for example, in an office setting, "temperature" is the primary indicator because it has the strongest impact on office efficiency).

[0118] The index-performance comparison table is a pre-defined table of correspondences between building physical indices and building physical performance (for example, "temperature" corresponds to "thermal comfort", "light intensity" corresponds to "visual comfort", and "noise level" corresponds to "acoustic comfort").

[0119] The primary performance is the building physical performance corresponding to the primary indicator (the indicator with the greatest influence) in the new sub-matrix (for example, in an office setting, "temperature" corresponds to "thermal comfort" as the primary performance).

[0120] The remaining primary indicators are those that affect the new submatrix, excluding the primary indicator with the greatest impact (for example, in an office setting, in addition to "temperature", "light intensity" and "noise level" are the remaining primary indicators).

[0121] The building simulation dimension is the direction of simulation analysis determined based on primary performance, secondary performance, and environmental data. It reflects the key performance aspects of the building that need to be simulated (for example, by combining the "thermal comfort" (primary performance) and "visual comfort" (secondary performance) of the office scenario with outdoor temperature data, we can set up the "energy consumption dimension" (related to air conditioning regulation of thermal comfort) and the "lighting dimension" (related to lighting regulation of visual comfort)).

[0122] The beneficial effects of the above technical solution are: by refining the analysis of user behavior needs, static and dynamic behavior data are transformed into quantifiable building physical performance, and simulation dimensions are precisely set in combination with environmental data, which significantly improves the matching degree between the simulation model and actual needs, and provides a precise analytical basis for subsequent building environment optimization.

[0123] This invention provides a method for building environment simulation and optimization, which sets building simulation dimensions based on a first performance and a second performance, and in conjunction with environmental data, including:

[0124] The environmental data is decomposed based on the attribute characteristics of each performance component to construct spatial entities;

[0125] After encoding and converting each spatial entity, it is mapped to an entity lookup table, and each spatial entity has a mapping occupancy coefficient in the entity lookup table based on the lookup area under each preset simulation dimension;

[0126]

[0127] Among them, R i represents the mapping occupancy coefficient of the reference region in the i-th preset simulation dimension; mi represents the number of spatial entities involved in the reference region in the i-th preset simulation dimension; M j Q represents the number of occupied units of the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j δ represents the weight of the performance corresponding to the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j represents the auxiliary function of the j-th spatial entity; ai represents the entity coefficient of the ith preset simulation dimension; Ni is the total number of elements in the reference region of the ith preset simulation dimension;

[0128] If the mapping occupancy coefficient is greater than the set coefficient of the corresponding reference area, then the corresponding preset simulation dimension will be regarded as the building simulation dimension.

[0129] Otherwise, the corresponding preset simulation dimension will be removed.

[0130] In this embodiment, the performance attributes are the core quantitative indicators of building physical performance (such as thermal comfort and visual comfort). For example, the attributes of thermal comfort are "temperature 24-26℃, humidity 40%-60%"; the attributes of visual comfort are "light intensity 300-500 lux, color temperature 3000-5000K".

[0131] Deconstructing environmental data involves filtering associated environmental parameters based on performance attributes and establishing a "performance-environment" mapping relationship. For example, thermal comfort is associated with "outdoor temperature and solar radiation" (which affects the indoor thermal environment); visual comfort is associated with "solar radiation and cloud cover" (which affects lighting).

[0132] A spatial entity is a virtual entity that integrates "environmental data + building spatial location" to represent the impact of the environment on a local area of ​​a building. For example, the west facade receives strong solar radiation from 14:00 to 16:00, so a "West Facade Thermal Radiation Spatial Entity" is constructed, including: location (west facade), time (14:00-16:00), and radiation intensity (500W / m²). 2 The east facade receives ample sunlight from 9:00 AM to 11:00 AM, creating a "solid lighting space on the east facade," including: location (east facade), time (9:00 AM - 11:00 AM), and radiation intensity (500 W / m²). 2 Specifically, this involves creating an "environmental entity" family in BIM software (such as Revit), associating it with an environmental data API (such as a weather station data interface), and marking the building surface area and its spatiotemporal attributes.

[0133] Encoding conversion assigns a unique identifier to spatial entities to facilitate data association and retrieval. For example, the entity encoding the thermal radiation on the west facade is SE_Heat_1416 (SE = west facade, Heat = thermal radiation, 1416 = time interval); the entity encoding the lighting on the east facade is EE_Light_0911 (EE = east facade, Light = lighting, 0911 = time interval).

[0134] In this embodiment, the value of ai ranges from 0.3 to 0.7, representing the proportion coefficient of the total number of units in the control area. It is obtained through calibration of historical projects. The occupied unit is defined as a building space grid of 1 square meter, which is automatically divided by BIM technology.

[0135] In this embodiment, Q j The value range is from 0 to 1, and Furthermore, the weights of the building performance corresponding to the spatial entities are pre-set.

[0136] The entity lookup table is a table relating storage space entity codes, performance types, and preset simulation dimensions, as shown in Table 1:

[0137] Table 1

[0138] Spatial Entity Coding Performance type Preset simulation dimensions Comparison Area (Building Zoning) SE_Heat_1416 Thermal comfort Energy consumption, temperature and humidity West facade office area EE_Light_0911 Visual comfort Lighting and energy consumption East facade conference room

[0139] Preset simulation dimensions are predefined potential simulation analysis directions (such as energy consumption, lighting, ventilation, acoustics, temperature and humidity, etc.).

[0140] The reference area is the analysis area in the building space that corresponds to the preset simulation dimension (such as the air conditioning load analysis of the energy consumption dimension corresponding to the "west facade office area").

[0141] The mapping occupancy coefficient quantifies the degree of influence of spatial entities on the preset simulation dimension.

[0142] In this embodiment, the weight of the performance corresponding to the j-th spatial entity is set by expert experience or data analysis.

[0143] The set coefficient is a preset threshold to determine whether Ri meets the retention conditions (e.g., the energy consumption dimension set coefficient is 0.4). Specifically, the threshold is calibrated through DOE (Design of Experiments) or historical project regression analysis (e.g., energy consumption ≥ 0.4, daylight ≥ 0.35).

[0144] The beneficial effects of the above technical solution are: by using environmental data, spatial entities, quantification coefficients, and dimensional filtering, the abstract building performance and environmental impact are transformed into calculable occupancy coefficients, thereby improving the accuracy of simulation dimension filtering. This avoids ineffective simulation of redundant dimensions and ensures that core performance (such as energy consumption and lighting) is simulated in a focused manner, thus achieving a precise match between building environment simulation and actual needs.

[0145] This invention provides a method for building environment simulation and optimization, which involves retrieving a matching reference association array and optimizing the spliced ​​submatrix using satisfaction coefficients to obtain a new submatrix, including:

[0146] Based on the theoretical mapping relationship between sub-demand scenarios and static features, a correlation strength tensor matrix is ​​constructed, where the matrix elements quantify the correlation strength between sub-demand scenarios and static features in numerical form.

[0147] The reference association arrays are prioritized based on the numerical values ​​of each element in the association strength tensor matrix, and reference association arrays with association strength exceeding a preset threshold are selected as the core optimization arrays.

[0148] The spliced ​​submatrix is ​​dimension-matched with the core optimization array. Mismatched dynamic feature dimensions are normalized to a spatial scale that is homologous with the static feature dimensions, generating a dimension-aligned intermediate matrix.

[0149] The elements in the dimension-aligned intermediate matrix are corrected based on the comprehensive correlation direction in the core optimization array. Specifically, if the comprehensive correlation direction is positive, then... Enhance the association weights between corresponding static and dynamic features, where G(fs) is a weight incremental function based on the positive association direction fs; mx is the satisfaction coefficient; and yx is a preset coefficient.

[0150] If the overall correlation direction is negative, then according to Weaken the corresponding association weights, where G(ff) is a weight decay function based on the negative direction of the comprehensive association ff;

[0151] Calculate the goodness of fit between the corrected dimension-aligned intermediate matrix and the theoretical mapping relationship. If the goodness of fit does not meet the preset standard, retrieve the secondary priority supplementary array from the reference associated array for secondary correction until the goodness of fit meets the requirements and obtain the new submatrix.

[0152] In this embodiment, G(fs) = 1 + k1·mx, where k1 is the attenuation coefficient and its value ranges from 0.05 to 0.1. Through experimental calibration, the default value is 0.1.

[0153] G(ff) = 1 - k2·mx, where k2 is the attenuation coefficient, and its value ranges from 0.05 to 0.1. Through experimental calibration, the default value is 0.1.

[0154] In this embodiment, the value of yx is 1 by default, which represents 100% of the theoretical satisfaction.

[0155] In this embodiment, the sub-requirement scenario is a set of user behaviors with a unified functional goal (such as "office scenario" including workstation work and meeting discussions, with the core goal of "completing work efficiently").

[0156] In this embodiment, the association strength tensor matrix is ​​a three-dimensional / high-dimensional matrix, and the element values ​​quantify the "degree of association between sub-demand scenarios and static features" (dimension: [scenario × static feature × association strength]). Taking the association strength tensor matrix of the office scenario (simplified to a two-dimensional example) as an example, as shown in Table 2:

[0157] Table 2

[0158] Static features Office hours Seat fixation Correlation strength value 0.8 0.7

[0159] In this embodiment, priority sorting is performed by sorting the reference association array according to the element values ​​of the association strength tensor matrix (such as 0.8 for office hours and 0.7 for seat fixity).

[0160] The preset threshold is a critical value for filtering the correlation strength (e.g., 0.6, which only retains arrays with correlation strength ≥ 0.6).

[0161] The core optimization array is a filtered array of highly correlated features. The more static features involved in a sub-requirement scenario, and the richer the dynamic features associated with each static feature, the larger the number of core optimization arrays may be. For example, in the "office scenario," the static features are [office hours, seat fixedity], and the dynamic features are [travel to / from the break room, distance traveled, document printing volume]:

[0162] Office hours Round trip to the tea room (association strength 0.7, exceeding threshold 0.6) → Core array 1;

[0163] Office hours File print count (association strength 0.65, exceeding threshold 0.6) → Core array 2;

[0164] Seat fixation Movement distance (association strength 0.55, not exceeding threshold 0.6) → non-core array; at this time, there are 2 core optimization arrays (because the association strength of the former two meets the standard).

[0165] A spliced ​​submatrix is ​​a matrix formed by splicing static state vectors (e.g., [8 hours, 0.9]) and dynamic state vectors (e.g., [5 times / day, 200 meters]) under the same sub-demand scenario. For example, in an office scenario, see Table 3:

[0166] Table 3

[0167]

[0168] Dimension matching aligns the feature dimensions of the spliced ​​submatrix with the core optimized array (if the core array only contains "number of round trips", then that column is retained, and "distance traveled" is temporarily stored).

[0169] Dynamic feature dimension normalization maps dynamic features (such as "5 round trips" and "200-meter travel distance") to the scale of static features (such as static features in "hours" or "proportions" and dynamic features in the [0,1] range through Min-Max normalization). For example, round trip normalization: (5-0) / (10-0) = 0.5 (assuming a theoretical maximum of 10 round trips); travel distance normalization: (200-0) / (500-0) = 0.4 (assuming a theoretical maximum travel distance of 500 meters).

[0170] The dimension-aligned intermediate matrix is ​​a normalized concatenated submatrix (dynamic and static features are on the same scale), as shown in Table 4:

[0171] Table 4

[0172]

[0173] The comprehensive correlation direction is a unified correlation trend (positive / negative) obtained by integrating and analyzing the correlation directions of multiple core optimization arrays. It is used to guide the correction of submatrices, and the comprehensive correlation direction is determined as follows:

[0174] Directional encoding: quantizes the associated direction (positive = +1, negative = -1);

[0175] Weighted summation: For each core optimization array, calculate the direction × correlation strength, and then sum them to obtain the total resultant force;

[0176] Judgment direction:

[0177] If the total resultant force > 0.2, the overall correlation direction is positive;

[0178] If the total resultant force is less than 0.2, the overall correlation direction is negative.

[0179] If the total resultant force is approximately 0 (e.g., absolute value ≤ 0.2), then the main trend is mapped by combining the theory of sub-demand scenarios.

[0180] The secondary priority supplementary array is an array whose association strength is lower than a preset threshold (e.g., 0.6), but can be supplemented and corrected (e.g., in an office scenario, the association array strength of the movement distance is 0.5, as a supplement).

[0181] The second correction involves adding supplementary arrays until the fit meets the target (e.g., MSE ≤ 0.05).

[0182] The beneficial effects of the above technical solution are: by tensor modeling, array filtering, dimension normalization, dynamic weight correction, and fitting verification, the fitting degree between the submatrix and the theoretical mapping is improved; by dynamically correcting and adapting to complex scenarios, a high-precision behavioral data foundation is provided for subsequent building physical performance analysis.

[0183] This invention provides a method for building environment simulation and optimization, which constructs individual error simulation diagrams corresponding to different dimensions of building simulation, including:

[0184] Based on the first alignment processing result, the trend term and fluctuation term in the time dimension and the regional distribution features in the spatial dimension are extracted, a feature dual matrix is ​​constructed, and the relative deviation, trend error and fluctuation error between the first process data and the set standard are calculated to form an error vector.

[0185] The error vector is mapped to the corresponding spatial region of the three-dimensional geometric model, and the error type is distinguished based on a preset legend to generate a corresponding individual error simulation diagram.

[0186] In this embodiment, the first alignment processing result is the comparison result of the first process data of a single dimension (such as energy consumption data of a certain hour segment) and the standard set for that dimension (such as energy consumption limit) after time and format unification (e.g., align the "hourly air conditioning energy consumption data" with the "10kWh per hour limit in the 'Energy Conservation Standard for Public Buildings'" on an hourly scale to form an "actual value-standard value" comparison table).

[0187] The trend in the time dimension refers to the long-term pattern of change (such as an upward, downward, or stable trend) of the first-process data in the time series.

[0188] The fluctuation term in the time dimension refers to the short-term random fluctuations of the first-process data around the trend term (such as the jumps in instantaneous energy consumption caused by the start-up and shutdown of equipment).

[0189] The regional distribution characteristics of the spatial dimension are the differences in the distribution of first-process data in different spatial areas of the building (e.g., the energy consumption distribution of rooms facing east, west, south, and north in an office building; rooms facing west generally have higher energy consumption than other areas due to strong solar radiation).

[0190] The eigenvalue dual matrix is ​​a two-dimensional matrix that integrates the time dimension (trend term, fluctuation term) and the spatial dimension (regional distribution characteristics). The rows represent the time step, the columns represent the spatial region, and the elements are the eigenvalues ​​of the corresponding spatiotemporal points, as shown in Table 5:

[0191] Table 5

[0192] Time / Region West-facing office East-facing office North-facing office 9:00 Trend Item 8kWh 6kWh 5kWh 9:00 Fluctuation Item +1kWh -0.5kWh 0kWh

[0193] The first process data is dynamic data generated in single-dimensional simulation (e.g., hourly air conditioning energy consumption: 9kWh at 9:00 and 10kWh at 10:00).

[0194] Setting standards are single-dimensional reference benchmarks (such as industry standards or design goals). For example, the energy consumption setting standard for air conditioners is "≤10kWh per hour".

[0195] The relative deviation is the percentage of the absolute difference between the data from the first process and the set standard to the standard value. For example, if the actual energy consumption is 12 kWh and the standard is 10 kWh, the relative deviation is (12-10) / 10 × 100% = 20%.

[0196] Trend error is the deviation between the trend term of the first process data and the trend term of the set standard (preset). (For example, the standard trend term is "energy consumption is stable at 10kWh from 9:00 to 18:00", and the actual trend term is "gradually rising to 15kWh". The trend error is the root mean square difference between the two: actual trend and standard trend.)

[0197] Fluctuation error is the deviation between the fluctuation item of the first process data and the fluctuation range (preset) allowed by the set standard (for example: the standard allows a fluctuation range of ±1kWh, the actual fluctuation item is +2kWh, and the fluctuation error is 1kWh).

[0198] The error vector is a three-dimensional vector composed of relative deviation, trend error, and fluctuation error, which quantifies the comprehensive error of a single dimension (e.g., [20%, 3kWh, 1kWh], corresponding to relative deviation, trend error, and fluctuation error, respectively).

[0199] The corresponding spatial region in the 3D geometric model is the specific space in the 3D building model associated with the error data (such as a west-facing office, an east-facing conference room, etc., corresponding to the room or area ID in the BIM model).

[0200] The default legend is a visual symbol system used to distinguish error types (for example, red indicates relative deviation >20%, yellow indicates 10%-20%, and green indicates <10%; dashed lines mark trend error areas, and solid lines mark fluctuation error areas).

[0201] Individual error simulation plots are dynamic graphs that visualize error vectors by spatial region, intuitively showing the spatiotemporal distribution of single-dimensional errors (for example, in the individual error simulation plot of energy consumption dimension, the west-facing office is highlighted in red (relative deviation 20%), and a dashed arrow indicates the upward direction of the trend error).

[0202] The beneficial effects of the above technical solution are: by separating time trends and fluctuation characteristics, constructing error vectors by combining spatial distribution, and then presenting individual error simulation diagrams in three-dimensional visualization, the abstract single-dimensional error data is transformed into an intuitive spatiotemporal correlation map of errors, thereby improving the accuracy of error positioning, providing precise targeting basis for subsequent anomaly intervention, and significantly improving the efficiency of single-dimensional performance optimization of buildings.

[0203] This invention provides a method for building environment simulation and optimization, which constructs a comprehensive error simulation diagram, including:

[0204] The global process data is sequentially aligned and compared with the setting standards of all single-dimensional simulations to generate the first global error simulation diagram.

[0205] The global process data is compared with the first process data under all building simulation dimensions to generate the first global deviation simulation diagram.

[0206] The first global deviation simulation map and the first global error simulation map are projected onto the three-dimensional geometric model of the building, respectively, to determine the building error form based on each building simulation dimension. When the building error form has an interactive form in other dimensions, auxiliary labels are set for the building error form based on the interactive form.

[0207] Based on the architectural error patterns and auxiliary labels under the corresponding architectural simulation dimensions, the corresponding dimension layers in the first global error simulation map are adjusted to obtain the comprehensive error simulation map.

[0208] In this embodiment, the sequential alignment comparison is performed dimensionally, aligning the global process data with the corresponding dimension's setting standard one scale at a time (the time axis is unified to the smallest time sequence unit, and the spatial axis is bound to the BIM component ID), and calculating the error, for example:

[0209] Energy consumption dimension: Global data 15kWh vs standard 10kWh → error +5kWh;

[0210] Temperature and humidity dimension: Global data 26℃ vs standard 24-26℃ → error 0℃ (critical value).

[0211] In this embodiment, the first global error simulation graph visualizes the error distribution of global data and standards in each dimension. The spatial dimension uses color gradients to map the error magnitude (e.g., red = energy consumption overrun, blue = abnormal temperature and humidity), and the time dimension uses dynamic frames to show the error evolution (e.g., one frame is updated every 15 minutes).

[0212] Dimensional alignment comparison unifies the spatiotemporal scale of global process data and single-dimensional first-dimensional process data (time axis aligned to the smallest temporal unit, spatial axis aligned to the same BIM partition), and calculates the deviation (deviation = global value - single-dimensional value), for example:

[0213] Energy consumption dimension: Global 15kWh vs single dimension 12kWh → deviation +3kWh;

[0214] Temperature and humidity dimensions: Global 26℃ vs. Single dimension 25℃ → Deviation +1℃.

[0215] The first global deviation simulation diagram visualizes the deviation distribution between global data and single-dimensional data. The spatial dimension uses a heat map to show the deviation magnitude (e.g., orange = large energy consumption deviation, green = small temperature and humidity deviation), and the time dimension uses a curve to show the deviation trend (e.g., energy consumption deviation increases over time).

[0216] Building error morphology refers to the distribution pattern of single-dimensional errors in building space (such as "band-like excess in west-facing rooms" in the energy consumption dimension, which appears as a continuous red area; and "corner point-like anomalies" in the temperature and humidity dimension, which appear as discrete red spots).

[0217] The interaction patterns in other dimensions are the coupling relationships of multi-dimensional errors (such as the overlap of energy consumption exceeding the standard area with temperature and humidity fluctuation area, forming an arrow transmission pattern of "energy consumption area → temperature and humidity area"; or the error in the noise dimension causes a compensatory error in the illumination dimension, forming a correlation curve of "noise↑ → illumination↑").

[0218] Auxiliary labels are symbols / text used to mark interaction relationships (such as arrows labeled "energy consumption → temperature and humidity", numbers labeled with correlation strength 0.8; or formulas labeled Δ energy consumption = 0.5·Δ temperature and humidity). For example, for the west-facing office area with excessive energy consumption, an arrow is superimposed pointing to the area with fluctuating temperature and humidity, labeled "correlation degree 0.8". Specifically, in the BIM model, the Dynamo script is used to detect the intersection of error areas in adjacent dimensions and automatically add geometric markers (arrows) + text labels.

[0219] The corresponding architectural error patterns in architectural simulation are spatial-temporal characteristics of single-dimensional error patterns (such as "continuous exceedance zone" in the energy consumption dimension and "instantaneous fluctuation point" in the temperature and humidity dimension).

[0220] Adjusting the corresponding dimension layers in the first global error simulation map is based on the spatial overlap of the error morphology and the correlation strength of the auxiliary labels. The visualization weight of the dimension layers is adjusted (e.g., increasing the color brightness of the coupled areas and reducing the transparency of the weakly correlated areas). For example, in the west-facing office energy consumption and temperature and humidity coupled areas (correlation degree 0.8), the red brightness of the energy consumption layer is increased by 50%, and the blue transparency of the temperature and humidity layer is reduced by 30% to highlight the interaction relationship.

[0221] The comprehensive error simulation diagram is a three-dimensional visualization that integrates multi-dimensional error distribution, interactive transmission relationships, and dynamic evolution trends. It supports dimension layer switching (such as hiding the lighting dimension and focusing on energy consumption-temperature and humidity coupling) and timeline backtracking (viewing historical error correlations). For example, a west-facing office can display "red highlight band (excessive energy consumption) → orange gradient area (temperature and humidity fluctuations)", with arrows and correlation labels overlaid, and the timeline plays the dynamic process of error expansion.

[0222] The beneficial effects of the above technical solution are as follows: by comparing two images (global error vs. global deviation) → morphological projection → interactive annotation → dimensional layer optimization, the spatial distribution, temporal evolution, and interactive transmission of multi-dimensional errors are integrated into a dynamic simulation image, which improves the efficiency of tracing the source of cross-dimensional anomalies, provides a global perspective and precise correlation decision-making basis for subsequent multi-dimensional collaborative optimization, and promotes the building environment simulation towards multi-dimensional collaboration.

[0223] This invention provides a method for building environment simulation and optimization, which synchronously evolves and intervenes in anomalies of the comprehensive error simulation map and the individual error simulation maps under each building simulation dimension in the smallest time series unit to generate a first optimization scheme, including:

[0224] Configure a unified timing reference clock for the integrated error simulation graph and each individual error simulation graph, extract the error feature vector of each smallest timing unit, and construct the error feature timing stream;

[0225] The internal dynamic correlation analysis and inter-time dynamic correlation analysis are performed on the time-series flow of the error characteristics to calculate the real-time correlation matrix between the comprehensive error and each individual error.

[0226] The error change rate prediction model, trained based on historical evolution data, generates a dynamic warning slope for each smallest time series unit.

[0227] When a dimensional error triggers a static threshold in a single error simulation graph, if the overall error does not exceed the dynamic warning slope, the corresponding dimension's error correction rule base will be automatically retrieved to generate and record the in-dimensional optimization instructions.

[0228] If the overall error exceeds the dynamic warning slope, then, based on the strong correlation errors existing in the real-time correlation matrix and the generation trajectory of the strong correlation errors, the parameter combination that minimizes the error collaboration is solved and recorded.

[0229] If a dimensional error exists in a single error simulation graph but does not trigger the static threshold, it is determined that there is no error in the corresponding dimension and recorded.

[0230] The first optimization scheme is generated based on the recorded results.

[0231] In this embodiment, a unified timing reference clock sets the same time reference for the integrated error simulation graph and all individual error simulation graphs (e.g., synchronization based on the NTP protocol, with timestamp accuracy ≤1ms), ensuring that the multiple graphs are fully aligned in the time dimension.

[0232] The error feature vector is a vector composed of the core features of the single-dimensional error / comprehensive error (such as deviation magnitude, trend slope, and fluctuation intensity) within each smallest time unit. For example, (energy consumption dimension): [20% (relative deviation), +5kW / h (trend slope), 2kW (fluctuation intensity)]; the comprehensive error vector needs to be superimposed with multi-dimensional features.

[0233] Error feature time series is a sequence of error feature vectors arranged in chronological order, forming a spatiotemporally correlated data stream. For example, the time series is [t0: vector 1, t1: vector 2, ..., tn: vector n], where each vector corresponds to 15 minutes of error features.

[0234] Internal dynamic correlation analysis is the correlation strength between the overall error and each individual error within the same time series (such as the synchronous change of energy consumption error and temperature and humidity error). For example, at time t0, the Pearson correlation coefficient between energy consumption error +20% and temperature and humidity error +1℃ is 0.8 (strong positive correlation).

[0235] Inter-time dynamic correlation analysis is the transmission relationship of error characteristics between different time series (such as the influence of energy consumption error at t0 on temperature and humidity error at t1). For example, if energy consumption at t0 exceeds the budget by 20%, it will cause the temperature and humidity at t1 to rise by 1°C, with a correlation delay of 1 time series unit.

[0236] The real-time correlation matrix is ​​a matrix whose elements are comprehensive errors. The correlation strength of each individual error (the fusion value of internal and inter-temporal correlations), with a dimension of (number of dimensions + 1) × (number of dimensions + 1) (including the comprehensive error), is shown in Table 6:

[0237] Table 6

[0238] Comprehensive error Energy consumption error Temperature and humidity error Lighting error Comprehensive error 1.0 0.8 0.7 0.3 Energy consumption error 0.8 1.0 0.6 0.2

[0239] In this embodiment, the error change rate prediction model is a prediction model (such as LSTM) trained based on historical evolution data (the error feature time series of the past 30 days), which predicts the error change rate of the next time series. The input is (time series window, number of features), and the output is the error change rate. The training set is historical data.

[0240] The dynamic warning slope is the upper limit of the allowable rate of error change for each smallest time series unit (predicted value × safety factor 1.2), preventing sudden error changes.

[0241] The static threshold is a preset upper limit of the error amplitude (such as energy consumption deviation ≥15%, temperature and humidity deviation ≥2℃), which triggers single-dimensional intervention and is preset in the JSON configuration file, such as {"energy consumption":0.15,"temperature and humidity":2}.

[0242] The error correction rule base contains preset correction strategies for each dimension (e.g., excessive energy consumption → reduce the air conditioner setting temperature by 1℃; excessive temperature and humidity → increase ventilation volume). For example, the rule base entry is {"energy consumption":{"action":"set_aircon_temp","param":-1}}.

[0243] Intra-dimensional optimization instructions are device control instructions generated by the rule base when a static threshold is triggered in a single dimension (such as "set the air conditioner temperature to 25℃").

[0244] Strongly correlated errors are error pairs with a correlation degree ≥ 0.7 (such as the strong correlation between energy consumption error and temperature and humidity error).

[0245] The generated trajectory is the path of strong correlation error over time (such as the synchronous trajectory of energy consumption error from +10% to +20% and temperature and humidity error from +0.5℃ to +1.5℃).

[0246] The parameter combination that minimizes error coordination is the optimal parameter for multi-dimensional coordinated adjustment (such as simultaneously adjusting the air conditioner temperature to -1℃ and the window opening angle to +15°), which minimizes the overall error.

[0247] The first optimization scheme is an executable solution that integrates single-dimensional instructions, collaborative parameters, and timing nodes, including:

[0248] Timing-based intervention points (e.g., adjusting air conditioning at 09:15, adjusting window opening at 09:30);

[0249] Dimensional adjustment parameters (e.g., energy consumption dimension -1°C, ventilation dimension +15°);

[0250] Prioritization (e.g., energy consumption regulation takes precedence over daylighting regulation).

[0251] The beneficial effects of the above technical solution are as follows: through time-series synchronization, correlation modeling, prediction-driven, and hierarchical intervention, a closed loop of dynamic monitoring → intelligent decision-making → precise control is constructed: single-dimensional intervention provides rapid response to local anomalies (efficiency improvement of 40%), multi-dimensional collaboration solves coupling problems (error convergence speed improvement of 50%), dynamic warning slope provides early warning of sudden change risks (false judgment rate reduction of 30%), and the final output first optimized solution achieves dynamic optimization with precise time-series, targeted dimensions, and coordinated parameters, providing an intelligent decision engine for the whole life cycle regulation of the building environment.

[0252] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for simulating and optimizing a building environment, characterized in that, include: Step 1: Collect architectural design drawings related to the built environment to construct a three-dimensional geometric model of the building. Combine the building's physical performance related to user behavior needs with the building simulation dimensions set by environmental data and integrate them into the three-dimensional geometric model of the building to obtain the first building model. Step 2: Perform single-dimensional and full-dimensional simulations on the first building model, and capture the first process data of each single-dimensional simulation and the global process data of the full-dimensional simulation. Step 3: Perform a first alignment process between the first process data and the setting standard of the single-dimensional simulation to construct a separate error simulation diagram for the corresponding building simulation dimension. At the same time, perform a second alignment process between the global process data and all first process data to construct a comprehensive error simulation diagram. Step 4: Synchronously evolve and intervene in anomalies of the comprehensive error simulation map and the individual error simulation map under each building simulation dimension in the smallest time sequence unit to generate the first optimization scheme, and perform initial optimization on the first building model and its actual application after initial optimization. Then, based on the difference between the actual running data and the simulation results after initial optimization, continue to optimize the model after initial optimization.

2. The building environment simulation and optimization method according to claim 1, characterized in that, Constructing a 3D geometric model of the building includes: Each structural component in the architectural design drawings is automatically extracted to generate a component data block with dimension annotations; The parametric components that match the component data blocks are matched from the component attribute database, and the corresponding parametric components are assembled using the design framework of the architectural design drawings to obtain a three-dimensional geometric model of the building.

3. The building environment simulation and optimization method according to claim 1, characterized in that, Determine the building's physical performance in relation to user behavior needs, and set building simulation dimensions based on environmental data, including: The user behavior requirements are analyzed to obtain a first state vector based on static behavior and a second state vector based on dynamic behavior. The first state vector and the second state vector are then extracted and concatenated for the same sub-requirement scenario to obtain a concatenated sub-matrix. The satisfaction coefficient of the corresponding sub-requirement scenario is determined. At the same time, the static behavior data and dynamic behavior data of the user in the building space are collected and scenario analysis is performed to obtain the dynamic features associated with each static feature. A reference association array for the corresponding static features is constructed, wherein the reference association array contains the association direction between the static features and each dynamic feature. Based on the theoretical mapping relationship between sub-demand scenarios and each static feature, the matching reference association array is retrieved, and the spliced ​​sub-matrix is ​​optimized by combining the satisfaction coefficient to obtain a new sub-matrix; Taking each preset building physics index as the entry point, interference analysis is performed on each new submatrix in turn to determine the index that affects the corresponding new submatrix, which is regarded as the first index, and the building physics performance of the corresponding new submatrix is ​​obtained based on the index-performance comparison table. The building physics performance corresponding to the first metric that is most affected by each new submatrix is ​​retained and regarded as the first performance. Meanwhile, the building physical performance corresponding to the remaining first index of each new sub-matrix is ​​intersected to obtain the second performance, wherein the second performance is the result of the building physical performance under the corresponding remaining first index appearing twice or more; Based on the first and second performance metrics, and combined with environmental data, the building simulation dimensions are set.

4. The building environment simulation and optimization method according to claim 3, characterized in that, Based on the first and second performance metrics, and in conjunction with environmental data, the building simulation dimensions are set, including: The environmental data is decomposed based on the attribute characteristics of each performance component to construct spatial entities; After encoding and converting each spatial entity, it is mapped to an entity lookup table, and each spatial entity has a mapping occupancy coefficient in the entity lookup table based on the lookup area under each preset simulation dimension; Among them, R i represents the mapping occupancy coefficient of the reference region in the i-th preset simulation dimension; mi represents the number of spatial entities involved in the reference region in the i-th preset simulation dimension; M j Q represents the number of occupied units of the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j δ represents the weight of the performance corresponding to the j-th spatial entity in the comparison region of the i-th preset simulation dimension; j represents the auxiliary function of the j-th spatial entity; ai represents the entity coefficient of the ith preset simulation dimension; Ni is the total number of elements in the reference region of the ith preset simulation dimension; If the mapping occupancy coefficient is greater than the set coefficient of the corresponding reference area, then the corresponding preset simulation dimension will be regarded as the building simulation dimension. Otherwise, the corresponding preset simulation dimension will be removed.

5. The method for simulating and optimizing the building environment according to claim 1, characterized in that, Retrieve the matching reference association array and optimize the concatenated submatrix using the satisfaction coefficients to obtain a new submatrix, including: Based on the theoretical mapping relationship between sub-demand scenarios and static features, a correlation strength tensor matrix is ​​constructed, where the matrix elements quantify the correlation strength between sub-demand scenarios and static features in numerical form. The reference association arrays are prioritized based on the numerical values ​​of each element in the association strength tensor matrix, and reference association arrays with association strength exceeding a preset threshold are selected as the core optimization arrays. The spliced ​​submatrix is ​​dimension-matched with the core optimization array. Mismatched dynamic feature dimensions are normalized to a spatial scale that is homologous with the static feature dimensions, generating a dimension-aligned intermediate matrix. The elements in the dimension-aligned intermediate matrix are corrected based on the comprehensive correlation direction in the core optimization array. Specifically, if the comprehensive correlation direction is positive, then... Enhance the association weights between corresponding static and dynamic features, where G(fs) is a weight incremental function based on the positive association direction fs; mx is the satisfaction coefficient; and yx is a preset coefficient. If the overall correlation direction is negative, then according to Weaken the corresponding association weights, where G(ff) is a weight decay function based on the negative direction of the comprehensive association ff; Calculate the goodness of fit between the corrected dimension-aligned intermediate matrix and the theoretical mapping relationship. If the goodness of fit does not meet the preset standard, retrieve the secondary priority supplementary array from the reference associated array for secondary correction until the goodness of fit meets the requirements and obtain the new submatrix.

6. The method for simulating and optimizing the building environment according to claim 1, characterized in that, Construct separate error simulation diagrams for the corresponding architectural simulation dimensions, including: Based on the first alignment processing result, the trend term and fluctuation term in the time dimension and the regional distribution features in the spatial dimension are extracted, a feature dual matrix is ​​constructed, and the relative deviation, trend error and fluctuation error between the first process data and the set standard are calculated to form an error vector. The error vector is mapped to the corresponding spatial region of the three-dimensional geometric model, and the error type is distinguished based on a preset legend to generate a corresponding individual error simulation diagram.

7. The method for simulating and optimizing the building environment according to claim 1, characterized in that, Construct a comprehensive error simulation diagram, including: The global process data is sequentially aligned and compared with the setting standards of all single-dimensional simulations to generate the first global error simulation diagram. The global process data is compared with the first process data under all building simulation dimensions to generate the first global deviation simulation diagram. The first global deviation simulation map and the first global error simulation map are projected onto the three-dimensional geometric model of the building, respectively, to determine the building error form based on each building simulation dimension. When the building error form has an interactive form in other dimensions, auxiliary labels are set for the building error form based on the interactive form. Based on the architectural error patterns and auxiliary labels under the corresponding architectural simulation dimensions, the corresponding dimension layers in the first global error simulation map are adjusted to obtain the comprehensive error simulation map.

8. The method for simulating and optimizing building environments according to claim 1, characterized in that, Synchronously evolve and intervene in anomalies in the comprehensive error simulation map and the individual error simulation maps under each building simulation dimension using the smallest time-series unit to generate the first optimization scheme, including: Configure a unified timing reference clock for the integrated error simulation graph and each individual error simulation graph, extract the error feature vector of each smallest timing unit, and construct the error feature timing stream; The internal dynamic correlation analysis and inter-time dynamic correlation analysis are performed on the time-series flow of the error characteristics to calculate the real-time correlation matrix between the comprehensive error and each individual error. The error change rate prediction model, trained based on historical evolution data, generates a dynamic warning slope for each smallest time series unit. When a dimensional error triggers a static threshold in a single error simulation graph, if the overall error does not exceed the dynamic warning slope, the corresponding dimension's error correction rule base will be automatically retrieved to generate and record the in-dimensional optimization instructions. If the overall error exceeds the dynamic warning slope, then, based on the strong correlation errors existing in the real-time correlation matrix and the generation trajectory of the strong correlation errors, the parameter combination that minimizes the error collaboration is solved and recorded. If a dimensional error exists in a single error simulation graph but does not trigger the static threshold, it is determined that there is no error in the corresponding dimension and recorded. The first optimization scheme is generated based on the recorded results.

Images