Method for predicting carbon emission of atrium shopping mall based on BP (Back Propagation) neural network

Abstract

The invention discloses an atrium shopping mall carbon emission prediction method based on a BP neural network, and relates to the technical field of carbon emission prediction. The invention aims to solve the problems of low prediction efficiency and low accuracy of the current shopping mall atrium carbon emission prediction method. The method comprises the following steps: acquiring a layout of a shopping mall, acquiring the type of a courtyard of the shopping mall, and establishing a shopping mall simulation model; acquiring carbon emission and photovoltaic power generation carbon reduction of the shopping mall simulation model; and building information data of the shopping mall simulation model is acquired, a training set is constructed by using the building information data of the shopping mall simulation model, the carbon emission of the shopping mall simulation model and the photovoltaic power generation carbon reduction, the BP neural network model is trained by using the training set, and the trained BP neural network model is used as a carbon emission prediction model. And acquiring to-be-predicted shopping mall building information data, and inputting the to-be-predicted shopping mall building information data into the carbon emission prediction model to obtain a carbon emission prediction value. The method is used for predicting carbon emission in the courtyard of the shopping mall.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission prediction, and in particular to a carbon emission prediction method for an atrium shopping mall based on a BP neural network. Background Art

[0002] As an important part of the construction industry, commercial buildings are characterized by large construction volume, rapid market growth, and high energy intensity. With the improvement of people's living standards, small shopping malls can no longer meet the needs of society, and the proportion of multifunctional integrated large shopping malls in commercial buildings has increased dramatically. As the most flexible spatial element in the shopping mall, the atrium space usually adopts a transparent and open design method to ensure that it has a good view and lighting conditions to alleviate the adverse psychological reactions of consumers who are in a closed space for a long time. In order to adapt to the requirements of the times, large shopping malls have developed into multifunctional comprehensive commercial centers that integrate leisure, exhibitions, catering, entertainment, and shopping, and the functions of the atrium have become more complex. Therefore, the operation of atrium space-related equipment and the maintenance of facility environment are bound to cause greater energy consumption, so carbon reduction in the atrium of large shopping malls has become a research focus in this field.

[0003] The current carbon emission prediction methods mainly use physical engine simulation methods and machine learning-based carbon emission prediction methods. However, the physical engine simulation method is time-consuming, labor-intensive, and highly professional, which leads to low efficiency in carbon emission prediction. The currently used machine learning-based carbon emission prediction method requires the collection of a large amount of carbon emission data in the early stage, while the carbon emission data collection cycle of the shopping mall atrium is long and the collection channels are difficult, resulting in low efficiency in carbon emission prediction. In addition, the current method has little consideration of the impact of the mall atrium morphology on carbon emissions, resulting in low accuracy in carbon emission prediction. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of low prediction efficiency and accuracy in the current shopping mall atrium carbon emission prediction method, and propose a shopping mall atrium carbon emission prediction method based on BP neural network.

[0005] A carbon emission prediction method for an atrium shopping mall based on a BP neural network comprises: obtaining building information data of a shopping mall to be predicted, and inputting the building information data of the shopping mall to be predicted into a carbon emission prediction model to obtain a carbon emission prediction value;

[0006] The shopping mall building information data to be predicted includes: the length and width of the building shape of the shopping mall to be predicted, the floor height, the number of floors, the total window area, the atrium size, the number of atriums, and the photovoltaic coverage rate;

[0007] The carbon emission forecast values ​​include: the total carbon emissions per cubic meter of the shopping mall building to be forecasted, the total carbon emissions per cubic meter of heating per cubic meter, the total carbon emissions per cubic meter of cooling per cubic meter, the total carbon emissions per cubic meter of lighting per cubic meter, the total photovoltaic power generation, and the carbon reduction per cubic meter of photovoltaic power generation per year;

[0008] The carbon emission prediction model is obtained by:

[0009] Step 1: Obtain the layout of the mall and the type of the mall atrium, and then establish a mall simulation model;

[0010] Step 2: Obtain the carbon emissions of the shopping mall simulation model;

[0011] Step 3: Obtain the carbon reduction of photovoltaic power generation in the shopping mall simulation model;

[0012] Step 4: Obtain the building information data of the shopping mall simulation model, use the building information data of the shopping mall simulation model, the carbon emissions of the shopping mall simulation model and the carbon reduction of photovoltaic power generation to construct a training set, use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model.

[0013] Furthermore, the atrium scale includes: atrium area ratio, skylight area ratio, atrium shape coefficient, connector height-to-width ratio, atrium concentration, atrium uniformity and vertical staggeredness, specifically;

[0014] The atrium area ratio is obtained by:

[0015]

[0016] Among them, S vj is the area of ​​the jth atrium, S b is the ground floor area of ​​the shopping mall, AAR is the atrium area ratio, and N is the number of atria in the shopping mall building to be predicted;

[0017] The skylight area ratio is obtained by the following method:

[0018]

[0019] Among them, SAR is the skylight area ratio, T aj is the light transmission area of ​​the jth atrium skylight, S t is the top interface area of ​​the shopping mall to be predicted;

[0020] The atrium shape coefficient is obtained by:

[0021]

[0022] Among them, F jis the surface area of ​​the jth atrium in contact with the mall, V j is the volume of the jth atrium;

[0023] The linker aspect ratio is obtained by:

[0024]

[0025] C h =C h1 +C h2

[0026]

[0027]

[0028] C w =C w1 +C w2

[0029]

[0030]

[0031] Among them, C h is the connector height, CSAR is the connector aspect ratio, C h1 is the percentage of the main atrium's corresponding height, C h2 The percentage of the height of the atrium, h a1 is the main atrium height, h a2 The height of the atrium is d w is the average width of the aisles on both sides of the atrium, C1 is the length of the connector, l a1 is the main atrium length, C w is the connector width, d a1 is the main atrium width, d a2 is the average width of the atrium, C w1 is the percentage of the corresponding width of the main atrium, C w2 The percentage of the corresponding width of the atrium;

[0032] The atrium concentration is obtained by:

[0033]

[0034]

[0035]

[0036] Among them, AI is the atrium concentration, is the observed average distance, is the expected mean distance, dj is the distance between the centroid of the jth atrium and the centroid of the atrium nearest to the centroid of the jth atrium;

[0037] The atrium uniformity is achieved by:

[0038]

[0039] Among them, V min is the volume of the smallest atrium, AU is the atrium uniformity;

[0040] The vertical staggered drop is obtained by:

[0041]

[0042] Where VSR is the vertical staggered ratio, h aj is the height of the jth atrium, h max is the maximum atrium height.

[0043] Furthermore, in step 1, the layout diagram of the shopping mall is obtained, and the type of each atrium is obtained, and then a shopping mall simulation model is established, specifically:

[0044] Step 11: Obtain layout diagrams of multiple shopping malls and obtain the type of atrium of each shopping mall;

[0045] The types of shopping mall atriums include: centralized atrium, chessboard atrium, radiant atrium, circular atrium, and linear atrium;

[0046] Step 1 and 2: Establish a benchmark model of the mall based on the mall layout diagram, and establish a mall simulation model using the size range of each element in the mall layout and the benchmark model;

[0047] A benchmark model of the mall is established based on the mall layout diagram, and then Latin hypercube sampling is used to conduct multiple samplings within the size range of each element in the mall layout to obtain multiple sets of sizes of each element in the mall layout, and a mall simulation model is established based on the benchmark model;

[0048] The dimensions of various elements in the shopping mall layout include: the length of the main atrium, the width of the main atrium, the length of the secondary atrium, the width of the secondary atrium, the length of the shopping mall mother body, the width of the shopping mall mother body, the closest distance from the secondary atrium to the long side of the shopping mall building, the area of ​​the atrium skylight, and the number of atria.

[0049] Furthermore, the step 2 of obtaining the carbon emissions of the shopping mall simulation model is specifically as follows:

[0050] C total =C heating +C cooling +C lighting

[0051]

[0052] Among them, C total is the annual total carbon emissions per cubic meter of the current shopping mall simulation model, C heating is the total annual heating carbon emissions per cubic meter of the current shopping mall simulation model, C cooling is the total carbon emissions per cubic meter of refrigeration per year for the current mall simulation model, C lighting is the annual total lighting carbon emissions per cubic meter of the current shopping mall simulation model, Q heating1 is the annual heating energy consumption of the atrium and corridor space of the current shopping mall simulation model, Q heating2 is the annual heating energy consumption of other spaces except atrium and corridor in the current shopping mall simulation model, f E is the average carbon emission factor of the current regional power grid, f C is the carbon emission factor of coal energy, V total It is the building volume of the current shopping mall simulation model.

[0053] Furthermore, the annual total carbon emissions per cubic meter of refrigeration of the current shopping mall simulation model is C cooling Specifically:

[0054]

[0055] Among them, Q cooling is the annual cooling energy consumption.

[0056] Furthermore, the annual lighting total carbon emissions per cubic meter of the current shopping mall simulation model is C lighting Specifically:

[0057]

[0058] Among them, Q lighting is the annual lighting energy consumption.

[0059] Furthermore, the step 3 of obtaining the carbon reduction amount of photovoltaic power generation of the shopping mall simulation model is specifically as follows:

[0060]

[0061] E pv =Q×S×η1×η

[0062] Among them, E pv is the annual carbon reduction per cubic meter of photovoltaic power generation in the shopping mall simulation model, Q is the annual total radiation of the photovoltaic system, S is the area of ​​the photovoltaic module, η1 is the conversion efficiency of the photovoltaic module, and η is the efficiency of the photovoltaic power generation system.

[0063] Furthermore, in step 4, the building information data of the shopping mall simulation model is obtained, and a training set is constructed using the building information data of the shopping mall simulation model, the carbon emissions of the shopping mall simulation model, and the carbon reduction of photovoltaic power generation. The BP neural network model is trained using the training set, and the trained BP neural network model is used as a carbon emission prediction model, specifically:

[0064] Step 41: Obtain building information data of the shopping mall simulation model, and form a training set with the building information data of the shopping mall simulation model, the carbon reduction of photovoltaic power generation, and the carbon emissions of the shopping mall simulation model;

[0065] The carbon emissions of the shopping mall simulation model include: the annual total carbon emissions per cubic meter of the shopping mall simulation model, the annual total carbon emissions per cubic meter of heating of the shopping mall simulation model, the annual total carbon emissions per cubic meter of refrigeration of the shopping mall simulation model, and the annual total carbon emissions per cubic meter of lighting of the shopping mall simulation model;

[0066] The building information data of the shopping mall simulation model includes: the length and width of the building form, the height of the storey, the number of floors, the total window area, the size of the atrium, the number of atriums, and the photovoltaic coverage rate corresponding to the shopping mall simulation model;

[0067] Step 42: Use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model;

[0068] The BP neural network model includes: an input layer, a hidden layer, and an output layer.

[0069] Furthermore, the number of neurons in the hidden layer of the BP neural network model is obtained by:

[0070]

[0071] Among them, N h is the number of neurons in the hidden layer, N s is the number of samples in the training set, α is a constant, N i' is the number of neurons in the input layer, N o is the number of neurons in the output layer.

[0072] Furthermore, the training set in step 4 trains the BP neural network model using the following loss function:

[0073]

[0074] Among them, k is the sample number in the training set, y k is the true value of the total carbon emissions per cubic meter per year of the kth sample, is the predicted value of total carbon emissions per cubic meter per year for the kth sample.

[0075] The beneficial effects of the present invention are:

[0076] The present invention proposes a method for predicting carbon emissions of atrium shopping malls based on a BP neural network model. First, the atrium type of the shopping mall is determined, and a benchmark model of the building scale is constructed; then, a simulation model of the shopping mall is established in batches through a parameterized method; then, the carbon emission data for BP neural network training is calculated; on this basis, the obtained training data set is input into the BP neural network model for training to obtain a trained BP neural network model; finally, the trained BP neural network model is used to predict carbon emissions of atrium shopping malls. The present invention constructs a model that can efficiently and accurately predict carbon emissions of large shopping malls with different atrium group forms, fully considering the factors that affect carbon emissions, thereby improving the accuracy of carbon emission prediction for shopping mall atriums. At the same time, the geometric model of the shopping mall trained by the present invention is obtained through simulation, and does not require the collection of a large amount of carbon emission data in the early stage, shortening the collection cycle and improving the efficiency of shopping mall carbon emission prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a flow chart of the present invention;

[0078] Figure 2(a) shows the centralized atrium layout;

[0079] Figure 2(b) shows the layout of the chessboard-style atrium;

[0080] Figure 2(c) shows the layout of the radial atrium;

[0081] Figure 2(d) is the layout diagram of the circular atrium;

[0082] Figure 2(e) is a layout diagram of a linear atrium;

[0083] Figure 3 Centralized atrium benchmark model;

[0084] Figure 4 It is a schematic diagram of the relevant element values ​​of the benchmark model;

[0085] Figure 5 This is the operation flow of the BP neural network. DETAILED DESCRIPTION

[0086] Specific implementation method 1: This implementation method is a BP neural network-based atrium shopping mall carbon emission prediction method. The specific process is as follows:

[0087] Obtaining building information data of the shopping mall to be predicted, and inputting the building information data of the shopping mall to be predicted into the carbon emission prediction model to obtain a carbon emission prediction value;

[0088] The shopping mall building information data to be predicted includes: the length and width of the building shape of the shopping mall to be predicted, the floor height, the number of floors, the total window area, the atrium size, the number of atriums, and the photovoltaic coverage rate;

[0089] The atrium scale includes: atrium area ratio, skylight area ratio, atrium shape coefficient, connector height-to-width ratio, atrium concentration, atrium uniformity and vertical staggeredness;

[0090] The atrium area ratio is obtained by:

[0091]

[0092] Among them, S vj is the area of ​​the jth atrium, S b is the ground floor area of ​​the shopping mall, AAR is the atrium area ratio, and N is the number of atria in the shopping mall building to be predicted;

[0093] The skylight area ratio is obtained by:

[0094]

[0095] Among them, SAR is the skylight area ratio, T aj is the light transmission area of ​​the jth atrium skylight, S t is the top interface area of ​​the shopping mall to be predicted;

[0096] The atrium shape coefficient is obtained by:

[0097]

[0098] Among them, F j is the surface area of ​​the jth atrium in contact with the mall, V j is the volume of the jth atrium;

[0099] The linker aspect ratio is obtained by:

[0100]

[0101] C h =C h1 +C h2

[0102]

[0103]

[0104] C w =C w1 +C w2

[0105]

[0106]

[0107] Among them, C h is the connector height, CSAR is the connector aspect ratio, C h1 is the percentage of the main atrium's corresponding height, C h2 The percentage of the height of the atrium, h a1 is the main atrium height, h a2 The height of the atrium is d w is the average width of the aisles on both sides of the atrium, C1 is the length of the connector, l a1 is the main atrium length, C w is the connector width, d a1 is the main atrium width, d a2 is the average width of the atrium, C w1 is the percentage of the corresponding width of the main atrium, C w2 The percentage of the corresponding width of the atrium;

[0108] The connector mentioned here only refers to the connector that can contain atrium monomers, and the connector length C1 refers to the length that includes all atrium monomers;

[0109] The atrium concentration is obtained by:

[0110]

[0111]

[0112]

[0113] Among them, AI is the atrium concentration, is the observed average distance, is the expected mean distance, d j is the distance between the centroid of the jth atrium and the centroid of the atrium nearest to the centroid of the jth atrium;

[0114] The atrium uniformity is achieved by:

[0115]

[0116] Among them, V min is the volume of the smallest atrium, AU is the atrium uniformity;

[0117] The vertical staggered drop is obtained by:

[0118]

[0119] Where VSR is the vertical staggered ratio, haj is the height of the jth atrium, h max is the maximum atrium height;

[0120] like Figure 1 As shown, the carbon emission prediction model is obtained by:

[0121] Step 1: Obtain the layout diagrams of different shopping malls, obtain the types of shopping mall atriums, and establish a shopping mall simulation model, specifically:

[0122] Step 11: Obtain layout diagrams of different shopping malls and obtain the type of atrium of each shopping mall;

[0123] like Figure 2(a)-Figure 2(e) The types of shopping mall atria include: centralized atria, chessboard atria, radiant atria, circular atria, and linear atria;

[0124] Step 1 and 2: Use Rhino software to establish a benchmark model of the current mall based on the mall layout diagram, and use the benchmark model and the size range of each element of the mall layout (through the analysis of 29 malls in cold regions with a total construction area of ​​no less than 100,000 m 2 The relevant parameters of the large shopping mall shape and atrium form are collected and sorted out), and the simulation model of the shopping mall is established using the grasshopper tool, such as Figure 4 As shown;

[0125] The Latin hypercube sampling method was used to uniformly sample the size range of each element of the mall layout to comprehensively examine the sample space of the elements, and at least 900 groups of sample data that met the requirements were obtained. Then, the grasshopper tool was used to parameterize and establish a simulation model.

[0126] The dimensions of the various elements of the mall layout include: the length of the main atrium la2, the width of the main atrium da2, the length of the secondary atrium la, the width of the secondary atrium da, the length of the mall matrix L, the width of the mall matrix D, the closest distance d1 from the secondary atrium to the long side of the building, the area of ​​the atrium skylight Sa, and the number of atria n (the sum of the number of main atria and secondary atria);

[0127] Step 2: Use the honeybee simulation calculation method to obtain the carbon emissions of each shopping mall simulation model, specifically:

[0128] C total =C heating +C cooling +C lighting

[0129]

[0130]

[0131]

[0132] Among them, C total is the annual total carbon emissions per cubic meter corresponding to the current shopping mall simulation model, C heating is the total annual carbon emissions per cubic meter of heating corresponding to the current shopping mall simulation model, C cooling is the total carbon emission per cubic meter of refrigeration per year corresponding to the current shopping mall simulation model, C lighting is the annual total lighting carbon emissions per cubic meter corresponding to the current shopping mall simulation model, Q heating1 is the annual heating energy consumption of the atrium and corridor space of the current shopping mall simulation model, Q heating2 is the annual heating energy consumption of spaces other than atriums and corridors, E is the average carbon emission factor of the current regional power grid, f C is the carbon emission factor of coal energy, V total is the current mall simulation model building volume, Q cooling is the annual cooling energy consumption, Q lighting is the annual lighting energy consumption.

[0133] In this step, Q heating1 , Q heating2 , Q cooling and Q lighting It is obtained through carbon emission simulation. All meteorological data used in the simulation are derived from the China Standard Meteorological Database (CSWD) and are selected according to the geographical location of the building under study.

[0134] Step 3: Obtain the carbon reduction C of photovoltaic power generation per cubic meter per year in the shopping mall simulation model building pv :

[0135]

[0136] E pv =Q×S×η1×η

[0137] Among them, f E is the average carbon emission factor of the current regional power grid, E pv is the annual power generation of the photovoltaic system (kWh), Q is the annual total radiation of the photovoltaic system (annual total radiation on the horizontal plane or annual total radiation on the inclined plane), S is the area of ​​the photovoltaic module, η1 is the conversion efficiency of the photovoltaic module, and η is the efficiency of the photovoltaic power generation system.

[0138] Due to the influence of latitude, the total amount of solar radiation received by different regions throughout the year will vary greatly. It needs to be checked according to the building conditions. Taking Harbin as an example, the value of Q obtained by querying the NASA database is 1379.6kWh / m 2The conversion efficiency of photovoltaic modules made of different materials also varies to a certain extent, and needs to be selected according to the actual design situation. According to the requirements of the "Standard Conditions for Photovoltaic Manufacturing Industry", the photovoltaic conversion efficiency of the battery module is set to be no less than 16%.

[0139] Step 4: Obtain the building information data of the shopping mall simulation model, use the carbon emissions of the shopping mall simulation model and the carbon reduction of photovoltaic power generation to build a training set, use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model, specifically:

[0140] Step 41: Obtain the building information data of the shopping mall simulation model, and form a training set with the building information data of the shopping mall simulation model, the carbon reduction of photovoltaic power generation, and the carbon emissions of the shopping mall simulation model;

[0141] The carbon emissions of the shopping mall simulation model include: the annual total carbon emissions per cubic meter of the shopping mall simulation model, the annual total carbon emissions per cubic meter of heating of the shopping mall simulation model, the annual total carbon emissions per cubic meter of refrigeration of the shopping mall simulation model, and the annual total carbon emissions per cubic meter of lighting of the shopping mall simulation model;

[0142] The building information data of the shopping mall simulation model includes: the length and width of the building form, the height of the storey, the number of floors, the total window area, the size of the atrium, the number of atriums, and the photovoltaic coverage rate corresponding to the shopping mall simulation model;

[0143] The length and width of the building form, storey height, number of floors, total window area, atrium size, and number of atria corresponding to the mall simulation model are the same as the method for obtaining the building information data of the mall to be tested;

[0144] Step 42: Use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model:

[0145] BP neural network is divided into two parts: information forward propagation and error back propagation. The operation process is as follows: Figure 5 As shown. The first task is to set the associated hyperparameters and the starting weights and biases, and then perform information flow in the order of input layer - hidden layer - output layer. Before establishing a BP neural network, it is necessary to configure the number of layers of the neural network, the number of neurons in each hidden layer, the activation function, and other hyperparameters.

[0146] The BP neural network model includes: an input layer, a hidden layer, and an output layer;

[0147] The number of neurons in the hidden layer is:

[0148]

[0149] Among them, N h is the number of neurons in the hidden layer, Ns is the number of samples in the training set, α∈[2,10] is a constant, N i' is the number of neurons in the input layer, N o is the number of neurons in the output layer.

[0150] The activation function is a Relu activation function;

[0151] The loss function is the mean square error MSE, specifically:

[0152]

[0153] Among them, k is the sample number in the training set, y k is the C of the kth sample total True value, is the C of the kth sample total Predicted value;

[0154] Step 4.3. Use the coefficient of determination R 2 It can be used to measure the goodness of fit of the carbon emission prediction model. 2 The closer the value is to 1, the better the fitting effect is. The calculation formula is as follows:

[0155]

[0156] in, is the C of the kth sample total Predicted mean.

[0157] Embodiment: In order to verify the beneficial effects of the present invention, the present invention carries out the following experiment:

[0158] S1. Select the atrium type of the prediction model according to the functional requirements, building scale and location of the building, and build a benchmark model of the building scale through Rhino. Figure 3 The central atrium is shown, with a total construction area of ​​about 120,000 m 2 , a length-to-width ratio close to 1:1, and 5 floors are used as the benchmark model for building scale.

[0159] S2. Use the Grasshopper tool to generate a parametric carbon emission simulation model and determine the value range of the first-level elements, including the length and width of the main atrium (la2, da2), the length and width of the secondary atrium (la, da), the length and width of the shopping mall mother body (L, D), the closest distance from the secondary atrium to the long side of the building (d1), the atrium skylight area (Sa) and the number of atria (n), a total of 9 items.

[0160] Subsequently, the Latin hypercube sampling method was used to perform uniform sampling within the value range of each factor to comprehensively examine the sample space of the factors, obtain at least 900 sets of sample data that meet the requirements, and then parameterize and establish a simulation model.

[0161] S3. Calculate the carbon emission data for BP neural network training through honeybee simulation:

[0162] The energy-carbon emission conversion method is determined according to the building area. Taking the Northeast region as an example, according to the survey, the small and medium-sized functional rooms in the shopping malls in the region use centralized heating, and the heating energy is mainly coal; the atrium corridor space uses central air conditioning for heating, which consumes the same energy as cooling and lighting, all of which are electric energy. Since the total building area of ​​the shopping mall will decrease as the atrium area increases, but the total volume of the shopping mall has not changed. Therefore, it is more appropriate to use carbon emissions per cubic meter as a measurement standard. Then calculate the annual carbon emissions per cubic meter of each part of the building operation stage and the annual carbon reduction per cubic meter of photovoltaic power generation.

[0163] In addition, due to the influence of latitude, the total amount of solar radiation received by different regions throughout the year will vary greatly. It needs to be obtained according to the building conditions. Taking Harbin as an example, through the query of NASA database, it can be known that the annual total radiation Q of the inclined surface (horizontal surface) is 1379.6kWh / m 2 The conversion efficiency of photovoltaic modules made of different materials also varies to a certain extent, and needs to be selected according to the actual design situation. According to the requirements of the "Standard Conditions for Photovoltaic Manufacturing Industry", the photovoltaic conversion efficiency of the battery module is set to be no less than 16%.

[0164] The building structure and equipment operation are set according to the relevant specifications [General Specification for Energy Conservation and Renewable Energy Utilization in Buildings: GB 55015-2021; Energy Conservation Design Standard for Public Buildings: GB 50189-2015]. Examples are shown in Tables 1 and 2.

[0165] f E According to the "Announcement of the Ministry of Ecology and Environment and the National Bureau of Statistics on the Release of the 2021 Electricity Carbon Dioxide Emission Factor", the average carbon emission factor of the power grid in the corresponding region is obtained. According to the relevant specifications [General Specification for Energy Efficiency and Renewable Energy Utilization in Buildings: GB 55015-2021; Energy Efficiency Design Standard for Public Buildings: GB 50189-2015], the building structure and equipment operation are set. The examples are shown in Tables 1 and 2 for simulation. The f in Northeast China E Take 0.6012kgCO2 / kwh; f C is the carbon emission factor of coal energy, taking 0.32kgCO2 / kwh;

[0166] Table 1 Building structure parameter settings

[0167]

[0168] Table 2 Equipment operation parameter settings

[0169]

[0170] S4, the length and width of the building shape, floor height, number of floors, total window area, atrium size, number of atria, photovoltaic coverage, C pv , C total , C heating , C cooling and C lighting A data set is formed and divided into a training set and a test set. The BP neural network model is trained and tested to obtain a carbon emission prediction model. The building information data of the atrium shopping mall to be predicted is predicted through the carbon emission prediction model module to obtain the carbon emission prediction result of the building.

[0171] In this embodiment, multiple BP neural network models with different numbers of hidden layers are set, and then each BP neural network model is trained using a training set, and then the trained BP neural network model is tested using a test set to obtain a deterministic coefficient of each trained BP neural network;

[0172] Then, the trained BP neural network model corresponding to the determination coefficient closest to 1 is used as the carbon emission prediction model.

[0173] According to the above experiments, the present invention achieves efficient and accurate prediction of carbon emissions through computer simulation combined with machine learning, thereby reducing the difficulty of prediction.

Claims

1. A method for predicting carbon emissions from atrium shopping malls based on BP neural network, characterized in that The specific process of the method is: obtaining the building information data of the shopping mall to be predicted, and inputting the building information data of the shopping mall to be predicted into the carbon emission prediction model to obtain the carbon emission prediction value; The building information data of the shopping mall to be predicted includes: the building length and width, floor height, number of floors, total window area, atrium size, number of atriums, and photovoltaic coverage rate of the shopping mall to be predicted; The carbon emission forecast values ​​include: the total carbon emissions per cubic meter of the shopping mall building to be forecasted, the total carbon emissions per cubic meter of heating per cubic meter, the total carbon emissions per cubic meter of cooling per cubic meter, the total carbon emissions per cubic meter of lighting per cubic meter, the total photovoltaic power generation, and the carbon reduction per cubic meter of photovoltaic power generation per year; The carbon emission prediction model is obtained by: Step 1: Obtain the layout of the mall and the type of the mall atrium, and then establish a mall simulation model; Step 2: Obtain the carbon emissions of the shopping mall simulation model; Step 3: Obtain the carbon reduction of photovoltaic power generation in the shopping mall simulation model; Step 4: Obtain the building information data of the shopping mall simulation model, use the building information data of the shopping mall simulation model, the carbon emissions of the shopping mall simulation model and the carbon reduction of photovoltaic power generation to construct a training set, use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model.

2. According to claim 1, a method for predicting carbon emissions of atrium shopping malls based on BP neural network is characterized by: The atrium scale includes: atrium area ratio, skylight area ratio, atrium shape coefficient, connector height-to-width ratio, atrium concentration, atrium uniformity and vertical staggeredness, specifically: The atrium area ratio is obtained by: Among them, S vj is the area of ​​the jth atrium, S b is the ground floor area of ​​the shopping mall, AAR is the atrium area ratio, and N is the number of atria in the shopping mall building to be predicted; The skylight area ratio is obtained by the following method: Among them, SAR is the skylight area ratio, T aj is the light transmission area of ​​the jth atrium skylight, S t is the top interface area of ​​the shopping mall to be predicted; The atrium shape coefficient is obtained by: Among them, F j is the surface area of ​​the jth atrium in contact with the mall, V j is the volume of the jth atrium; The linker aspect ratio is obtained by: C h =C h1 +C h2 C w =C w1 +C w2 Among them, C h is the connector height, CSAR is the connector aspect ratio, C h1 is the percentage of the main atrium's corresponding height, C h2 The percentage of the height of the atrium, h a1 is the main atrium height, h a2 The height of the atrium is d w is the average width of the aisles on both sides of the atrium, C1 is the length of the connector, l a1 is the main atrium length, C w is the connector width, d a1 is the main atrium width, d a2 is the average width of the atrium, C w1 is the percentage of the corresponding width of the main atrium, C w2 The percentage of the corresponding width of the atrium; The atrium concentration is obtained by: Among them, AI is the atrium concentration, is the observed average distance, is the expected mean distance, d j is the distance between the centroid of the jth atrium and the centroid of the atrium nearest to the centroid of the jth atrium; The atrium uniformity is achieved by: Among them, V min is the volume of the smallest atrium, AU is the atrium uniformity; The vertical staggered drop is obtained by: Where VSR is the vertical staggered ratio, h aj is the height of the jth atrium, h max is the maximum atrium height.

3. The carbon emission prediction method for atrium shopping malls based on BP neural network according to claim 2 is characterized by: The step 1 is to obtain the layout of the shopping mall, obtain the type of each atrium, and then establish a shopping mall simulation model, specifically: Step 11: Obtain layout diagrams of multiple shopping malls and obtain the type of atrium of each shopping mall; The types of shopping mall atriums include: centralized atrium, chessboard atrium, radiant atrium, circular atrium, and linear atrium; Step 1 and 2: Establish a benchmark model of the mall based on the mall layout diagram, and establish a mall simulation model using the size range of each element in the mall layout and the benchmark model; A benchmark model of the mall is established based on the mall layout diagram, and then Latin hypercube sampling is used to conduct multiple samplings within the size range of each element in the mall layout to obtain multiple sets of sizes of each element in the mall layout, and a mall simulation model is established based on the benchmark model; The dimensions of various elements in the shopping mall layout include: the length of the main atrium, the width of the main atrium, the length of the secondary atrium, the width of the secondary atrium, the length of the shopping mall mother body, the width of the shopping mall mother body, the closest distance from the secondary atrium to the long side of the shopping mall building, the area of ​​the atrium skylight, and the number of atria.

4. The carbon emission prediction method for atrium shopping malls based on BP neural network according to claim 3 is characterized by: The step 2 of obtaining the carbon emissions of the shopping mall simulation model is specifically as follows: C total =C heating +C cooling +C lighting Among them, C total is the annual total carbon emissions per cubic meter of the current shopping mall simulation model, C heating is the total annual heating carbon emissions per cubic meter of the current shopping mall simulation model, C cooling is the total carbon emissions per cubic meter of refrigeration per year for the current shopping mall simulation model, C lighting is the annual total lighting carbon emissions per cubic meter of the current shopping mall simulation model, Q heating1 is the annual heating energy consumption of the atrium and corridor space of the current shopping mall simulation model, Q heating2 is the annual heating energy consumption of other spaces except atrium and corridor in the current shopping mall simulation model, f E is the average carbon emission factor of the current regional power grid, f C is the carbon emission factor of coal energy, V total It is the building volume of the current shopping mall simulation model.

5. The method for predicting carbon emissions of atrium shopping malls based on BP neural network according to claim 4 is characterized by: The total annual carbon emissions per cubic meter of refrigeration in the current shopping mall simulation model is C cooling Specifically: Among them, Q cooling is the annual cooling energy consumption.

6. The method for predicting carbon emissions of atrium shopping malls based on BP neural network according to claim 5 is characterized by: The total annual lighting carbon emissions per cubic meter of the current shopping mall simulation model are C lighting Specifically: Among them, Q lighting is the annual lighting energy consumption.

7. The carbon emission prediction method for atrium shopping malls based on BP neural network according to claim 6 is characterized by: The step 3 of obtaining the carbon reduction amount of photovoltaic power generation of the shopping mall simulation model is specifically as follows: AND pv =Q×S×η1×η Among them, E pv is the annual carbon reduction per cubic meter of photovoltaic power generation in the shopping mall simulation model, Q is the annual total radiation of the photovoltaic system, S is the area of ​​the photovoltaic module, η1 is the conversion efficiency of the photovoltaic module, and η is the efficiency of the photovoltaic power generation system.

8. The method for predicting carbon emissions of atrium shopping malls based on BP neural network according to claim 7 is characterized by: The step 4 includes obtaining the building information data of the shopping mall simulation model, constructing a training set using the building information data of the shopping mall simulation model, the carbon emissions of the shopping mall simulation model, and the carbon reduction of photovoltaic power generation, training the BP neural network model using the training set, and using the trained BP neural network model as a carbon emission prediction model, specifically: Step 41: Obtain building information data of the shopping mall simulation model, and form a training set with the building information data of the shopping mall simulation model, the carbon reduction of photovoltaic power generation, and the carbon emissions of the shopping mall simulation model; The carbon emissions of the shopping mall simulation model include: the annual total carbon emissions per cubic meter of the shopping mall simulation model, the annual total carbon emissions per cubic meter of heating of the shopping mall simulation model, the annual total carbon emissions per cubic meter of refrigeration of the shopping mall simulation model, and the annual total carbon emissions per cubic meter of lighting of the shopping mall simulation model; The building information data of the shopping mall simulation model includes: the length and width of the building form, the height of the storey, the number of floors, the total window area, the size of the atrium, the number of atriums, and the photovoltaic coverage rate corresponding to the shopping mall simulation model; Step 42: Use the training set to train the BP neural network model, and use the trained BP neural network model as a carbon emission prediction model; The BP neural network model includes: an input layer, a hidden layer, and an output layer.

9. The method for predicting carbon emissions of atrium shopping malls based on BP neural network according to claim 8 is characterized by: The number of neurons in the hidden layer of the BP neural network model is obtained in the following way: Among them, N h is the number of neurons in the hidden layer, N s is the number of samples in the training set, α is a constant, N i' is the number of neurons in the input layer, N o is the number of neurons in the output layer.

10. The carbon emission prediction method of atrium shopping mall based on BP neural network according to claim 9 is characterized by: The training set in step 4 trains the BP neural network model using the following loss function: Among them, k is the sample number in the training set, y k is the true value of the total carbon emissions per cubic meter per year of the kth sample, is the predicted value of total carbon emissions per cubic meter per year for the kth sample.

Images