A method and system for predicting regional building carbon emissions

Abstract

The present invention is a method and system for predicting regional building carbon emissions. The process of the prediction method is: first, classify the information of individual buildings into static features and dynamic features according to whether they will change dynamically over time; second, input the similar components after decomposition and the mutation components after fusion of the dynamic features of the individual buildings into the dynamic spatiotemporal intervention model, which can realize the automatic update of the weighted graph and finally obtain the predicted dynamic features; finally, for different building types at different regional levels in the region, select the regional building static feature data set and the regional building predicted dynamic feature data set to construct a functional relationship with carbon emissions, and obtain the building carbon emission prediction results of the community, block and urban area after calculation. The present invention improves the precision and accuracy of regional building carbon emission prediction.

Description

Technical Field

[0001] The present invention belongs to the technical field of regional low-carbon planning, and in particular relates to a method and system for predicting regional building carbon emissions. Background Art

[0002] In regional low-carbon planning, buildings, as the most concentrated places for human activities, consume a lot of natural resources and have a huge impact on the ecological environment. In order to reduce carbon emissions in the regional construction sector, the first thing is to conduct a comprehensive and detailed carbon emission accounting and make accurate regional building carbon emission forecasts to provide guidance for the subsequent formulation of carbon reduction paths and the proposal of planning schemes.

[0003] Existing building carbon emission prediction methods mainly focus on cities and single buildings, and there is little involvement in regional-level building carbon emission prediction. In addition, most traditional building carbon emission prediction methods are limited by the characteristics of statistical methods, and have problems such as poor regression stability and low accuracy. They lack the ability of self-learning and cannot achieve continuous improvement of results. In recent years, deep learning has been widely used in the field of prediction research, usually including machine learning prediction, time series prediction, and image prediction.

[0004] Machine learning prediction models (such as support vector machines and BP artificial neural networks) have achieved many results in carbon emissions and energy consumption prediction in the construction industry. For example, the invention patent with publication number CN112906956A discloses a city energy consumption prediction method that combines BIM and CIM. This method establishes a time series model based on historical building energy consumption data and the correlation between influencing factors and time. The time series set is used as the input of the BP model to obtain the building energy consumption prediction power, which effectively reduces the prediction error caused by the nonlinear relationship between various influencing factors and building energy consumption.

[0005] Time series prediction mainly includes recurrent neural network (RNN) and long short-term memory network (LSTM), while image prediction mainly refers to graph convolutional neural network (GCN). At present, some scholars have combined the two to propose spatiotemporal graph convolutional neural network (STGCN). This model fully considers the spatial structure and correlation between data and has higher accuracy than machine learning prediction models. The invention patent with publication number CN116108983A proposes a method and device for predicting urban carbon emissions based on collaborative spatiotemporal convolutional network. This method constructs a multi-perspective heterogeneous graph based on urban carbon emissions and traffic density, and makes collaborative predictions of urban carbon emissions based on this multi-perspective heterogeneous graph, thereby improving the accuracy of urban carbon emissions prediction.

[0006] Regions can be divided into communities, blocks and urban areas according to their size. In regional building low-carbon planning, communities usually include residential buildings and the public areas around them; blocks and urban areas include residential buildings, office buildings, hotel buildings, commercial buildings and public areas. Regions at different levels have different building types and cannot be generalized when calculating.

[0007] Regional building carbon emission prediction usually involves a variety of building information. The building age, building space distribution, equipment quantity and equipment power usage usually do not change; the building area, number of buildings, number of users, and equipment operation time will change dynamically, resulting in seasonal and periodic changes in building carbon emissions in a certain month, day or time. If you want to accurately predict regional building carbon emissions, you cannot ignore the potential static and dynamic characteristics of the building information itself. The limitations of the above methods affect the prediction accuracy and overall scientific control of regional building carbon emissions. Summary of the invention

[0008] The present invention aims to overcome the shortcomings of the prior art, and in particular proposes a method and system for predicting regional building carbon emissions. First, the information of individual buildings is classified into static features and dynamic features based on whether they will change dynamically over time; secondly, the similar components after decomposition and the mutation components after fusion of the dynamic features of the individual buildings are input into the dynamic spatiotemporal intervention model, which can realize the automatic update of the weighted graph and finally obtain the predicted dynamic features; finally, for different building types at different regional levels in the region, the regional building static feature data set and the regional building predicted dynamic feature data set are selected to construct a functional relationship with carbon emissions, and the building carbon emission prediction results of the community, block and urban area are obtained after calculation, which improves the accuracy and precision of the regional building carbon emission prediction.

[0009] To achieve the above object, the technical solution of the present invention is:

[0010] In a first aspect, the present invention provides a method for predicting regional building carbon emissions, the method comprising the following steps:

[0011] Step S1: Acquire the single building information of different building types in the area to be studied; classify the area to be studied into residential areas, blocks and urban areas according to the regional level; the building types include residential buildings, office buildings, commercial buildings, hotel buildings and public areas; the single building information of each building type includes building related information, energy consumption related information and personnel and equipment information;

[0012] Step S2: Classify the static features and dynamic features of the individual building information of all building types according to whether they will change dynamically over time, and obtain the regional building static feature data set and the regional building dynamic feature data set; then decompose each dynamic feature of the individual building into similar components and mutation components at different scales, and fuse the mutation components of each dynamic feature at different scales to obtain the mutation components of each dynamic feature after fusion;

[0013] Step S3: Construct a regional building carbon emission prediction model and predict:

[0014] The regional building carbon emission prediction model includes a community building carbon emission calculation module, a block building carbon emission calculation module, an urban building carbon emission calculation module and a plurality of parallel dynamic spatiotemporal intervention models; the input of each dynamic spatiotemporal intervention model is a similar component of a dynamic feature of a single building and a fused mutation component, and the number of dynamic spatiotemporal intervention models is the same as the number of dynamic features; each dynamic spatiotemporal intervention model outputs a predicted dynamic feature; the outputs of all dynamic spatiotemporal intervention models are combined to obtain a regional building predicted dynamic feature data set;

[0015] According to the different regional levels of the area to be studied and the building types of individual buildings contained in the regional levels, the regional building prediction dynamic feature dataset and the regional building static feature dataset are respectively input into the community building carbon emission calculation module, the block building carbon emission calculation module and the urban building carbon emission calculation module to obtain the carbon emission prediction results of the community, block and urban buildings.

[0016] Furthermore, the dynamic spatiotemporal intervention model includes a dynamic spatiotemporal graph learning module, a spatial domain convolution module and a temporal domain convolution module;

[0017] The dynamic spatiotemporal graph learning module is used to construct a similar component weighted graph and a mutation component weighted graph for the similar component and the fused mutation component respectively according to the spatial relationship of the individual buildings in the area; the graph intervention parameters are set in the process so that the weighted graph can be automatically updated;

[0018] The spatial domain convolution module is used to fuse the similarity component weighted graph and the mutation component weighted graph to obtain the fused dynamic features;

[0019] The time domain convolution module is used to perform long-term feature capture on the fused dynamic features output by the spatial domain convolution module, and output the predicted dynamic features of a certain dynamic feature in the regional building dynamic feature data set.

[0020] Furthermore, the types of buildings included in the block and urban area are residential buildings, office buildings, commercial buildings, hotel buildings and public areas; the types of buildings included in the community are residential buildings and public areas;

[0021] The community building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings and public areas in the community in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions;

[0022] The block building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings, public areas, office buildings, commercial buildings and hotel buildings in the block in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions;

[0023] The urban building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings, public areas, office buildings, commercial buildings and hotel buildings in the urban area in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions.

[0024] Furthermore, in step S1, the building related information includes the building area , Construction Year and building location ; Energy consumption related information including gas historical data , power consumption historical data , heating historical data , water consumption historical data ; Personnel and equipment information including historical data of the number of users , Historical occupancy data 、Number of devices , Equipment power usage , Equipment running time ;in, , is the total number of buildings of different building types in the area; the equipment has multiple types according to the actual situation. In this case, the third digit can be introduced after the subscript n, and the information of different types of equipment can be represented by 1, 2, 3, etc., such as , Respectively represent the number of two different types of equipment.

[0025] In step S2, the static feature is the building age , Geographical location of the building 、Number of devices and equipment power usage ; The dynamic feature is the building area , Gas historical data , power consumption historical data , heating historical data , water consumption historical data , historical data of number of users , Historical occupancy data and equipment operating time ; The static and dynamic characteristics of the equipment increase as the number of equipment types increases.

[0026] Furthermore, the regional building static feature dataset It is expressed as:

[0027] .

[0028] in, for The elements in .

[0029] Further, in step S3, the regional building prediction dynamic feature data set It is expressed as:

[0030] ;

[0031] in, for The elements in ;

[0032] Dynamic features , , , , , , , The corresponding predicted dynamic characteristics.

[0033] The functional relationship between the regional building static feature dataset and the regional building prediction dynamic feature dataset and carbon emissions for each type of single building in the region is constructed, which is expressed as:

[0034] ,

[0035] ,

[0036] ,

[0037] ,

[0038] ,

[0039] In the formula, Carbon emission prediction results for residential single buildings; , is the total number of residential buildings in the area; is the functional calculation formula for carbon emissions of residential single buildings, It is a residential building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a residential building;

[0040] Carbon emission prediction results for single office buildings; ; is the total number of office buildings in the area; is the functional calculation formula for carbon emissions of a single office building, It is a feature recognition factor for an office building, which is used to identify and obtain the static features and predict the dynamic features corresponding to a certain office building;

[0041] Carbon emission prediction results for commercial single buildings; ; is the total number of commercial single buildings in the area; is the functional calculation formula for carbon emissions of commercial single buildings, It is a commercial single building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a commercial single building;

[0042] Carbon emission prediction results for a single hotel building; ; is the total number of hotel buildings in the area; is the functional calculation formula for the carbon emissions of a single hotel building, It is a hotel single building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a hotel single building;

[0043] Predict carbon emissions for public area buildings; ; is the total number of public areas in the region; is the functional calculation formula for carbon emissions of buildings in public areas, It is a public area building feature identification factor, which is used to identify the static features corresponding to a public area and predict the dynamic features.

[0044] Furthermore, the prediction results of carbon emissions from residential buildings are expressed as:

[0045] ,

[0046] In the formula, For the Prediction results of building carbon emissions in each community; It is the sum of the predicted carbon emissions of individual residential buildings in the community; For the Prediction results of building carbon emissions in public areas of each community;

[0047] The prediction results of urban building carbon emissions are expressed as:

[0048] ,

[0049] In the formula, Prediction results of carbon emissions from buildings in urban areas; is the sum of the predicted results of carbon emissions from buildings in urban areas. , is the number of subdistricts in the urban area; It is the sum of the predicted carbon emissions of office buildings in the urban area; It is the sum of the predicted carbon emissions of hotel buildings in the urban area; It is the sum of the predicted carbon emissions of commercial buildings in the urban area.

[0050] In addition, the carbon emission prediction results of block buildings The calculation method of is the same as the calculation method of urban building carbon emission prediction results. , , , The corresponding modifications are the number of residential communities in the block, the number of office buildings in the block, the number of hotel buildings in the block, and the number of commercial buildings in the block.

[0051] In a second aspect, the present invention provides a regional building carbon emission prediction system, which performs the steps of the prediction method, and the prediction system comprises:

[0052] An information acquisition module is used to acquire the individual building information of different building types in the area and input it into the information decomposition module;

[0053] The information decomposition module is used to classify the static and dynamic features of the individual building information of all building types according to whether they will change dynamically over time after receiving the individual building information from the information acquisition module, and obtain the regional building static feature data set and the regional building dynamic feature data set; then decompose each dynamic feature of the individual building into similar components and mutation components at different scales, and fuse the mutation components of each dynamic feature of the individual building at different scales to obtain the mutation components of each dynamic feature after fusion; the information decomposition module is connected to the carbon emission prediction module;

[0054] The carbon emission prediction module includes a community building carbon emission calculation module, a block building carbon emission calculation module, an urban building carbon emission calculation module and several parallel dynamic spatiotemporal intervention models; it is used to obtain a regional building prediction dynamic feature data set using similar components and fused mutation components, and then combined with a regional building static feature data set, according to different building types in the regional hierarchy, to obtain the carbon emission prediction results for communities, blocks and urban buildings.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention fully considers the static and dynamic changes of single building information over time, and classifies it into static features and dynamic features. According to the regional level of the area to be studied and the building types of single buildings contained in the regional level, the regional building prediction dynamic feature data set and the regional building static feature data set are respectively input into the community building carbon emission calculation module, the block building carbon emission calculation module and the urban building carbon emission calculation module to obtain the carbon emission prediction results of the community, block and urban buildings, so that the regional building carbon emission prediction accuracy is higher.

[0057] The dynamic spatiotemporal intervention model proposed in the present invention is improved and optimized on the basis of the existing spatiotemporal graph convolutional neural network (STGCN). It can directly learn the dynamic spatiotemporal dependencies across time slices, take into account the static and dynamic change characteristics of building information, fully consider the dynamic changes of the weighted graph, and set the graph intervention parameters so that the weighted graph can be automatically updated, which fully conforms to the dynamic change trend of the information of a single building. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of a method for predicting regional building carbon emissions of the present invention.

[0059] Figure 2 It is a relationship diagram of the regional hierarchical division and the corresponding building types of the present invention and a diagram of the obtained single building information content.

[0060] Figure 3 It is a flow chart of step S2 in the present invention.

[0061] Figure 4 It is a flow chart of step S3 in the present invention.

[0062] Figure 5 It is a flow chart of the dynamic spatiotemporal intervention model in the present invention. DETAILED DESCRIPTION

[0063] In order to better understand the technical content of the present invention, specific examples are given and described as follows with reference to the accompanying drawings.

[0064] Various aspects of the present invention are described herein with reference to the accompanying drawings, which provide many illustrative examples. The examples of the present invention are not limited to those described in the accompanying drawings. In addition, some of the contents disclosed herein may be used alone or in combination with other contents disclosed herein.

[0065] The present invention proposes a method for predicting regional building carbon emissions. Figure 1 As shown, the following steps are included:

[0066] Step S1: Obtain the individual building information of different building types in the area to be studied; classify the area to be studied into residential areas, blocks and urban areas according to the regional level; the building types include residential buildings, office buildings, commercial buildings, hotel buildings and public areas; the individual building information of each building type includes building-related information, energy consumption-related information, personnel and equipment information, such as Figure 2 As shown, building-related information includes building area , Construction Year and building location ; Energy consumption related information including gas historical data , power consumption historical data , heating historical data , water consumption historical data ; Personnel and equipment information including historical data of the number of users , Historical occupancy data 、Number of devices , Equipment power usage , Equipment running time , where the amount of building related information, energy consumption related information, and personnel and equipment information can be obtained according to actual conditions.

[0067] Assume that the area to be studied is a community There are several residential buildings and a public area in the community. Since the building area and construction age of all residential buildings in the community are similar, the building area of ​​the acquired single building information is , Construction Year , Historical occupancy data The impact on carbon emissions is not considered. In addition, the equipment that needs to be considered in this community is mainly elevators and street lights to predict the community The carbon emissions of buildings are used as an example to introduce it.

[0068] S11: Community The types of buildings included are residential buildings and public areas. All residential buildings and public areas in the express, , For the community The total number of residential buildings and public areas. The individual building information includes the building's geographical location , Gas historical data , power consumption historical data , heating historical data , water consumption historical data , Historical data of number of users 、Number of devices (The equipment includes elevators and street lights, and their quantities are recorded as the number of elevators 、Number of street lights ), equipment power usage (The equipment includes elevators and street lamps, and their power usage is recorded as elevator power usage , Street light power )、Equipment running time (The equipment includes elevators and street lights, and their running time is recorded as elevator running time , Street light operation time ), in this embodiment, a total of 12 single building information are considered.

[0069] Step S2: Classify the static features and dynamic features of the individual building information of all building types according to whether they will change dynamically over time, and obtain the regional building static feature dataset and the regional building dynamic feature dataset; then decompose each dynamic feature of the individual building into similar components and mutation components at different scales, and fuse the mutation components of each dynamic feature at different scales to obtain the mutation components of each dynamic feature after fusion, such as Figure 3 shown.

[0070] S21: Community Take the example to introduce that the cell Static features include building location 、Number of elevators 、Number of street lights , Elevator power usage , Street light power The cell Dynamic features include gas history data , power consumption historical data , heating historical data , water consumption historical data , Historical data of number of users , Elevator running time , Street light operation time .

[0071] Furthermore, the regional building static feature dataset It is expressed as:

[0072] .

[0073] in, for In this embodiment, .

[0074] S22: Further, the community Each dynamic feature of each single building in the image is decomposed into similar components and mutation components at different scales using wavelet transform, using historical data of the number of people. As an example, the decomposition methods of other dynamic features are the same as Consistent:

[0075] ,

[0076] ,

[0077] In the formula, represents the similarity component of the historical data of the number of users obtained by decomposition, ; It represents the mutation component of the historical data of the number of users obtained by decomposition; Represents the number of users of a single building in a historical time series. is a real number, Indicates the length of the time series; represents the total transformation scale; Indicates the number of users of a single building in a historical time series. The mutation component under the transformation scale, .

[0078] The mutation component of the historical data of the number of users obtained by decomposition It contains mutation components at different transformation scales, which are fused according to the following formula:

[0079] ,

[0080] ,

[0081] In the formula, represents the fusion result, is the mutation component of the historical data of the number of users after fusion; express Activation function; yes product; and represents the weight matrix; and represents the bias matrix; It is channel-by-channel convolution; Represents point-wise convolution; , represents a learnable parameter; and Respectively Functions and Activation function.

[0082] Step S3: Construct a regional building carbon emission prediction model and predict:

[0083] The regional building carbon emission prediction model includes several parallel dynamic spatiotemporal intervention models, community building carbon emission calculation modules, block building carbon emission calculation modules and urban building carbon emission calculation modules; the input of each dynamic spatiotemporal intervention model is the similarity component of a dynamic feature of a single building and the fused mutation component, and the number of dynamic spatiotemporal intervention models is the same as the number of dynamic features of the single building; each dynamic spatiotemporal intervention model outputs predicted dynamic features; the outputs of all dynamic spatiotemporal intervention models are combined to obtain a regional building predicted dynamic feature data set; according to the different regional levels of the area to be studied and the building types of the single buildings contained in the regional levels, the regional building predicted dynamic feature data set and the regional building static feature data set are respectively input into the community building carbon emission calculation module, the block building carbon emission calculation module and the urban building carbon emission calculation module to obtain the carbon emission prediction results of the community, block and urban buildings, such as Figure 4 shown.

[0084] By community Take this as an example for introduction.

[0085] S31: The dynamic spatiotemporal intervention model includes a dynamic spatiotemporal graph learning module, a spatial domain convolution module and a temporal domain convolution module, such as Figure 5 shown.

[0086] S311: The dynamic spatiotemporal graph learning module is based on the community The spatial relationship of the individual buildings in the network is analyzed, and a similarity component weighted graph and a mutation component weighted graph are constructed for the similarity component and the fused mutation component respectively; the graph intervention parameters are set in the process so that the weighted graph can be automatically updated.

[0087] S3111: Based on historical data of the number of users For example, the community The weighted graph representation of the similarity component of the historical data of the number of users is:

[0088] ,

[0089] ,

[0090] in, A weighted graph representing similar components of historical data on the number of occupants of a single building; represents the identity matrix; Represents the node correlation distance metric adjacency matrix; Represents the adjacency matrix of non-adjacent node similarity measures; represents the adaptive adjacency matrix; , Both represent graph intervention parameters; N represents the number of nodes. The number of nodes in the weighted graph is consistent with the total number of all buildings in the area; Embedding dimension for nodes; Chebyshev First-order information aggregation; is the feature dimension of the output vector; represents the learnable embedding matrix generated by random initialization; and Respectively Functions and Activation function.

[0091] S3112: Based on historical data of the number of users For example, the community The weighted graph of the mutation component of the historical data of the number of users is expressed as:

[0092] ,

[0093] ,

[0094] in, A weighted graph representing the mutation component of the historical data of the number of occupants of a single building; represents the identity matrix; Represents the node correlation distance metric adjacency matrix; Represents the adjacency matrix of non-adjacent node similarity measures; represents a dynamic adjacency matrix; , All represent figure intervention parameters; Embedding dimension for nodes; Chebyshev First-order information aggregation; is the feature dimension of the output vector; represents the dynamic update factor; and Respectively Functions and Activation function.

[0095] S312: The spatial domain convolution module includes Activation function and Product, the weighted graph of similar components is used After activation function processing, it is used together with the mutation component weighted graph The product is used to achieve fusion, and the fused dynamic features are obtained by using the historical data of the number of users. For example, it is expressed as:

[0096] ,

[0097] In the formula, express Activation function; yes product; Indicates the dynamic characteristics of the fused historical data of the number of users.

[0098] S313: The time domain convolution module is used to perform long-term feature capture on the fused dynamic features output by the spatial domain convolution module, and output the predicted dynamic features of a certain dynamic feature, so as to use the historical data of the number of people. For example, it is expressed as:

[0099] ,

[0100] In the formula, represents causal convolution; represents a learnable parameter; Represents the dynamic characteristics of the number of people using a single building.

[0101] S32: Further, the outputs of all dynamic spatiotemporal intervention models are merged to obtain a regional building prediction dynamic feature dataset , expressed as:

[0102] ,

[0103] in, for In this embodiment, ,

[0104] S33 Further, the community building carbon emission calculation module is used to calculate the community building carbon emission according to the functional relationship between the characteristics of residential buildings and public areas in the regional building prediction dynamic feature data set and the regional building static feature data set and carbon emissions. Carbon buildings emit carbon.

[0105] ,

[0106] ,

[0107] In the formula, For the community Carbon emission prediction results of residential single buildings; , For the community The total number of residential buildings in is the functional calculation formula for carbon emissions of residential single buildings; It is a characteristic identification factor of residential buildings, used to identify the residential area. The static characteristics and predicted dynamic characteristics of a residential building in the community Carbon emission activities related to residential buildings mainly include gas, electricity consumption, and heating. Main options for residential single buildings , , Construct functional relationships; , , They are the carbon emission factors for gas, electricity and heat respectively.

[0108] ,

[0109] ,

[0110] For the community Carbon emission prediction results of public area buildings; The function calculation formula for carbon emissions of buildings in public areas; It is a public area building feature identification factor used to identify the community The static features and predicted dynamic features corresponding to the public areas in the community The water pump room, elevators and street lights involved in water use activities are all installed in public areas. Main selection of public areas , , , , , , , Construct functional relationships; is the carbon emission factor for water use, is the carbon emission factor of electricity.

[0111] Furthermore, the community The building carbon emission prediction results are as follows:

[0112] ,

[0113] In the formula, For the community The predicted results of building carbon emissions are For the community The sum of the predicted results of carbon emissions from residential buildings.

[0114] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A method for predicting regional building carbon emissions, It is characterized in that The prediction method comprises the following steps: Step S1: Acquire the single building information of different building types in the area to be studied; classify the area to be studied into residential areas, blocks and urban areas according to the regional level; the building types include residential buildings, office buildings, commercial buildings, hotel buildings and public areas; the single building information of each building type includes building related information, energy consumption related information and personnel and equipment information; Step S2: Classify the static features and dynamic features of the individual building information of all building types according to whether they will change dynamically over time, and obtain the regional building static feature data set and the regional building dynamic feature data set; then decompose each dynamic feature of the individual building into similar components and mutation components at different scales, and fuse the mutation components of each dynamic feature at different scales to obtain the mutation components of each dynamic feature after fusion; Step S3: Construct a regional building carbon emission prediction model and predict: The regional building carbon emission prediction model includes a community building carbon emission calculation module, a block building carbon emission calculation module, an urban building carbon emission calculation module and a plurality of parallel dynamic spatiotemporal intervention models; the input of each dynamic spatiotemporal intervention model is a similar component of a dynamic feature of a single building and a fused mutation component, and the number of dynamic spatiotemporal intervention models is the same as the number of dynamic features; each dynamic spatiotemporal intervention model outputs a predicted dynamic feature; the outputs of all dynamic spatiotemporal intervention models are combined to obtain a regional building predicted dynamic feature data set; According to the different regional levels of the area to be studied and the building types of individual buildings contained in the regional levels, the regional building prediction dynamic feature data set and the regional building static feature data set are respectively input into the community building carbon emission calculation module, the block building carbon emission calculation module and the urban building carbon emission calculation module to obtain the carbon emission prediction results of the community, block and urban buildings; The dynamic spatiotemporal intervention model includes a dynamic spatiotemporal graph learning module, a spatial domain convolution module and a temporal domain convolution module; The dynamic spatiotemporal graph learning module is used to construct a similar component weighted graph and a mutation component weighted graph for the similar component and the fused mutation component respectively according to the spatial relationship of the individual buildings in the area; the graph intervention parameters are set in the process so that the weighted graph can be automatically updated; The spatial domain convolution module is used to fuse the similarity component weighted graph and the mutation component weighted graph to obtain the fused dynamic features; The time domain convolution module is used to perform long-term feature capture on the fused dynamic features output by the spatial domain convolution module, and output the predicted dynamic features of a certain dynamic feature in the regional building dynamic feature data set.

2. According to the method for predicting regional building carbon emissions in claim 1, It is characterized in that The types of buildings included in the block and urban area are residential buildings, office buildings, commercial buildings, hotel buildings and public areas; the types of buildings included in the community are residential buildings and public areas; The community building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings and public areas in the community in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions; The block building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings, public areas, office buildings, commercial buildings and hotel buildings in the block in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions; The urban building carbon emission calculation module is used to construct a functional relationship based on the characteristics of residential buildings, public areas, office buildings, commercial buildings and hotel buildings in the urban area in the regional building prediction dynamic feature data set and the regional building static feature data set, and calculate carbon emissions.

3. The regional building carbon emission prediction method according to claim 1, It is characterized in that In step S1, the building related information includes the building area A 1n 、Building Year A 2n and building location A 3n ; Energy consumption related information includes gas historical data B 1n 、Power consumption historical data B 2n 、Heating historical data B 3n 、Historical data of water consumption B 4n ; Personnel and equipment information including historical data of the number of users C 1n , Occupancy rate historical data C 2n 、Number of devices C 3n 、Equipment power consumption C 4n , Equipment running time C 5n ; Where n = 1, 2, ..., N, N is the total number of buildings of different building types in the area.

4. The regional building carbon emission prediction method according to claim 3, It is characterized in that In step S2, the static feature is the building age A 2n 、Building Geographical Location A 3n 、Number of devices C 3n And the equipment power C 4n ; The dynamic feature is the building area A 1n 、Gas historical data B 1n 、Power consumption historical data B 2n 、Historical data of heating B 3n 、Historical data of water consumption B 4n 、Historical data of number of users C 1n , Occupancy rate historical data C 2n and equipment operating time C 5n ; Regional building static feature dataset X S It is expressed as: X S ={X S1 ,X S2 ,...,X Sn ,...,X SN }; Among them, X Sn For X S The elements in X Sn ={A 2n , A 3n , C 3n , C 4n }.

5. The regional building carbon emission prediction method according to claim 4, It is characterized in that In step S3, the regional building prediction dynamic feature dataset X D It is expressed as: X D ={X D1 ,X D2 ,...,X Dn ,...,X DN }; Among them, X Dn For X D The elements in X Dn ={A 1n ″,B 1n ″,B 2n ″,B 3n ″,B 4n ″,C 1n ″,C 2n ″,C 5n ″}; A 1n ″,B 1n ″,B 2n ″,B 3n ″,B 4n ″,C 1n ″,C 2n ″,C 5n ″ are dynamic features A 1n , B 1n , B 2n , B 3n , B 4n , C 1n , C 2n , C 5n The corresponding predicted dynamic characteristics.

6. The regional building carbon emission prediction method according to claim 5, It is characterized in that The functional relationship between the regional building static feature dataset and the regional building prediction dynamic feature dataset and carbon emissions for each type of single building in the region is constructed, which is expressed as: In the formula, ZZ h is the carbon emission prediction result of a single residential building; h = 1, 2, ..., H, H is the total number of single residential buildings in the area; is the functional calculation formula for carbon emissions of residential single buildings, It is a residential building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a residential building; BG b is the carbon emission prediction result of a single office building; b = 1, 2, ..., B; B is the total number of single office buildings in the region; is the functional calculation formula for carbon emissions of a single office building, It is a feature recognition factor for an office building, which is used to identify and obtain the static features and predict the dynamic features corresponding to a certain office building; SY s is the carbon emission prediction result of a commercial single building; s = 1, 2, ..., S; S is the total number of commercial single buildings in the area; is the functional calculation formula for carbon emissions of commercial single buildings, It is a commercial single building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a commercial single building; JD i is the carbon emission prediction result of the hotel single building; i = 1, 2, ..., I; I is the total number of hotel single buildings in the area; is the functional calculation formula for the carbon emissions of a single hotel building, It is a hotel single building feature identification factor, which is used to identify the static features and predict the dynamic features corresponding to a hotel single building; GQ g is the carbon emission prediction result of buildings in public areas; g = 1, 2, ..., G; G is the total number of public areas in the region; is the functional calculation formula for carbon emissions of buildings in public areas, It is a public area building feature identification factor, which is used to identify the static features corresponding to a public area and predict the dynamic features.

7. The regional building carbon emission prediction method according to claim 6, It is characterized in that The prediction results of carbon emissions from residential buildings are expressed as follows: In the formula, XQ a is the predicted result of building carbon emissions in the ath community; It is the sum of the predicted results of carbon emissions of individual residential buildings in the community; GQ a The predicted results of building carbon emissions in the public area of ​​the ath community; The prediction results of urban building carbon emissions are expressed as: Where CQ is the predicted result of building carbon emissions in urban areas; is the sum of the predicted results of carbon emissions from buildings in the urban area, a=1, 2, ..., A, where A is the number of residential areas in the urban area; It is the sum of the predicted carbon emissions of office buildings in the urban area; It is the sum of the predicted carbon emissions of hotel buildings in the urban area; It is the sum of the predicted carbon emissions of commercial buildings in the urban area.

8. A regional building carbon emission prediction system, It is characterized in that The steps of executing the prediction method according to any one of claims 1 to 7, wherein the prediction system comprises: An information acquisition module is used to acquire the individual building information of different building types in the area and input it into the information decomposition module; The information decomposition module is used to classify the static and dynamic features of the individual building information of all building types according to whether they will change dynamically over time after receiving the individual building information from the information acquisition module, and obtain the regional building static feature data set and the regional building dynamic feature data set; then decompose each dynamic feature of the individual building into similar components and mutation components at different scales, and fuse the mutation components of each dynamic feature of the individual building at different scales to obtain the mutation components of each dynamic feature after fusion; the information decomposition module is connected to the carbon emission prediction module; The carbon emission prediction module includes a community building carbon emission calculation module, a block building carbon emission calculation module, an urban building carbon emission calculation module and several parallel dynamic spatiotemporal intervention models; it is used to obtain a regional building prediction dynamic feature data set using similar components and fused mutation components, and then combined with a regional building static feature data set, according to different building types in the regional hierarchy, to obtain the carbon emission prediction results for communities, blocks and urban buildings.

Images