Building carbon emission real-time online monitoring system

Abstract

The invention provides a real-time on-line building carbon emission monitoring system, and relates to the field of carbon emission monitoring, the real-time on-line building carbon emission monitoring system comprises a detection box body, a controller, a negative pressure fan and a sensor, the negative pressure fan and the sensor are installed in the detection box body, the top wall of the detection box body is provided with an air inlet, an air inlet pipe is installed in the detection box body, and the air inlet end of the air inlet pipe is provided with a filter screen; exhaust pipe openings are formed in the two sides of the detection box body, the sensor is used for collecting air environment data, and the controller is used for adjusting the air opening angle of the air inlet pipe by analyzing the collected air environment data. In this way, by dynamically adjusting the angle of the air opening of the air inlet pipe, quick response to environment changes and more accurate monitoring data collection can be achieved.

Description

Technical Field

[0001] The present application relates to the field of carbon emission monitoring, and more specifically, to a real-time online monitoring system for building carbon emissions. Background Art

[0002] Carbon emissions refer to greenhouse gas emissions generated during the product life cycle, including carbon dioxide, methane and nitrous oxide, etc., which are mainly caused by human activities. Excessive emissions lead to global warming, extreme weather, rising sea levels and ecosystem damage. The construction industry is one of the important sources of global energy consumption and carbon emissions. Therefore, monitoring carbon emissions in buildings is of great practical significance.

[0003] In this regard, the existing patent CN118937578A proposes a real-time online monitoring system for building carbon emissions. Specifically, an air inlet is provided on the top wall of the detection box, and an air inlet pipe with an adjustable air inlet angle is movably installed inside, and a filter is embedded in the air inlet end of the air inlet pipe. Exhaust pipe openings are provided on both side walls of the detection box. When the negative pressure fan rotates, the outside air is drawn through the filter into the detection box for detection, and the purified air is then discharged from the exhaust pipe opening.

[0004] The angle adjustment method of the air inlet duct in the above-mentioned existing patents only relies on mechanical linkage, that is, the angle adjustment of the air inlet duct is achieved by overdriving the motor, pulley transmission and differential rotation of large and small gears. However, this fixed mode adjustment has some limitations. Specifically, on the one hand, this mechanism cannot be dynamically adjusted according to real-time environmental changes, which may result in that in some cases, the air inlet duct may not be able to effectively capture the best air sample, resulting in incomplete or inaccurate monitoring data. On the other hand, the mechanical transmission system requires a certain amount of time to complete the angle adjustment, and the response speed is slow. This delay makes it difficult for the system to respond quickly to rapid changes in the environment, such as sudden changes in wind direction or sudden deterioration of air quality, thereby affecting monitoring efficiency and accuracy.

[0005] Therefore, a real-time online monitoring solution for building carbon emissions is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a real-time online monitoring system for building carbon emissions.

[0007] According to one aspect of the present application, a real-time online monitoring system for building carbon emissions is provided, which includes: a detection box, a negative pressure fan and a sensor installed inside the detection box, wherein the top wall of the detection box is provided with an air inlet, an air inlet pipe is provided in the air inlet, a filter is installed at the air inlet end of the air inlet pipe, exhaust pipe openings are provided on both side walls of the detection box, and the sensor is used to collect air environment data, wherein the real-time online monitoring system for building carbon emissions also includes a controller, and the controller is used to adjust the air outlet angle of the air inlet pipe; Wherein, the controller comprises: An air environment data receiving module, used to receive the air environment data collected by the sensor; An air environment data time series collaborative analysis module is used to perform a core time series collaborative analysis on the air environment data based on the carbon emission main variable and the environmental data auxiliary variable to obtain a recommended air outlet angle value; The air inlet angle adjustment module is used to input the recommended air inlet angle value and the real-time air inlet angle of the air inlet duct into the PLC controller to obtain a control instruction, and the control instruction is used to adjust the air inlet angle of the air inlet duct.

[0008] Compared with the prior art, the real-time online monitoring system for building carbon emissions provided by the present application includes a detection box, a controller, a negative pressure fan and a sensor installed inside the detection box, an air inlet is provided on the top wall of the detection box, an air inlet pipe is installed inside, a filter is provided at the air inlet end of the air inlet pipe, and exhaust pipe openings are provided on both sides of the detection box, wherein the sensor is used to collect air environment data, and the controller is used to adjust the air outlet angle of the air inlet pipe by analyzing the collected air environment data. In this way, by dynamically adjusting the air outlet angle of the air inlet pipe, rapid response to environmental changes and more accurate monitoring data collection can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 It is a block diagram of a controller in a real-time online monitoring system for building carbon emissions according to an embodiment of the present application.

[0010] Figure 2 It is a block diagram of the air environment data time series collaborative analysis module in the real-time online monitoring system of building carbon emissions according to an embodiment of the present application.

[0011] Figure 3Schematic diagram of data flow of the air environment data timing collaborative analysis module in the real-time online monitoring system for building carbon emissions according to an embodiment of the present application.

[0012] Figure 4 It is a block diagram of a timing depth coordination unit in a real-time online monitoring system for building carbon emissions according to an embodiment of the present application. DETAILED DESCRIPTION

[0013] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0014] Carbon emissions refer to greenhouse gas emissions generated during the product life cycle, including carbon dioxide, methane and nitrous oxide, which are mainly caused by human activities. Excessive emissions will lead to global warming, further causing problems such as extreme weather, rising sea levels and ecosystem damage. As one of the important sources of global energy consumption and carbon emissions, it is of great practical significance to monitor the carbon emissions of the construction industry.

[0015] The existing patent CN118937578A proposes a real-time online monitoring system for building carbon emissions. Specifically, an air inlet is provided on the top of the detection box, and an adjustable air inlet pipe is installed inside, and a filter is embedded in the air inlet end. Exhaust pipe openings are provided on both sides of the detection box. Through the operation of the negative pressure fan, the outside air enters the box through the filter for detection, and is discharged from the exhaust pipe opening after purification.

[0016] However, the angle adjustment method of the air inlet pipe in the patent mainly relies on mechanical linkage, that is, it is achieved through the drive motor, pulley transmission and differential rotation of large and small gears. This fixed mode has certain limitations: on the one hand, it cannot be dynamically adjusted according to real-time environmental changes, which may cause the air inlet pipe to fail to effectively capture the best air sample, thereby affecting the comprehensiveness and accuracy of the monitoring data; on the other hand, the mechanical transmission system takes a certain amount of time to adjust the angle, and the response speed is slow, making it difficult to adapt to rapid changes in the environment in a timely manner, such as sudden changes in wind direction or sudden deterioration of air quality, thereby reducing monitoring efficiency and accuracy.

[0017] Based on the existing patent CN118937578A, this application proposes a real-time online monitoring system for building carbon emissions, including: a detection box, a negative pressure fan installed inside the detection box, and a sensor, wherein the top wall of the detection box is provided with an air inlet, an air inlet pipe is provided inside the air inlet, a filter is installed at the air inlet end of the air inlet pipe, and exhaust pipe openings are provided on both side walls of the detection box. The sensor is used to collect air environment data, and also includes a controller, which is used to adjust the air outlet angle of the air inlet pipe. Specifically, the negative pressure fan is installed inside the detection box, fixedly plugged with the drive shaft, and driven by the drive motor to rotate the drive shaft to drive its operation. When the negative pressure fan rotates, it draws air from the outside of the monitoring area, so that the outside air can enter the detection box, provide air samples to be tested for detecting building carbon emissions, and ensure that the monitoring system can continuously obtain outside air for detection. The air inlet arranged on the top wall of the detection box is the entrance channel for the outside air to enter the detection box. It provides an installation position for the air inlet pipe, and at the same time plays a certain supporting and positioning role for the air inlet pipe, ensuring that the air inlet pipe can work stably, so that the outside air can enter the detection box in an orderly manner through the air inlet pipe. The air inlet pipe is installed inside the air inlet. It not only guides the outside air into the detection box, but also can collect air in multiple directions, so that the air from different directions can enter the detection box and be monitored in real time, thereby improving the accuracy of the monitoring results. The filter is embedded and installed at the air inlet end of the air inlet pipe to filter impurities and particulate matter in the outside air. These impurities and particulate matter are prevented from entering the detection box, preventing them from causing pollution and damage to components such as sensors, ensuring the normal operation of the detection instrument, ensuring that the detection results are not interfered by impurities, and improving the reliability of the monitoring data. The main function of the exhaust pipe openings located on the two side walls of the detection box is to discharge the air after the detection inside the detection box is completed. In this way, the flow and pressure balance of the air inside the detection box can be maintained, ensuring that new outside air can continue to smoothly enter the detection box for detection, so that the monitoring system can operate continuously and stably.

[0018] Correspondingly, in the controller, the technical concept of the present application is to receive the air environment data (wind speed value, wind direction value, ambient temperature value, ambient humidity value and carbon emission concentration value time series) collected by the sensor, and use the data analysis and encoding method based on deep learning to group the air environment data into variables to obtain the time series of carbon emission concentration value and the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value}, and then, perform time series encoding on these two time series, so as to intelligently obtain the recommended air outlet angle value according to the time series synergy feature between the encoded carbon emission auxiliary variables and the core time series deep synergy representation between the time series correlation feature of the carbon emission concentration, and input it into the PLC controller with the real-time air outlet angle to adaptively adjust the air outlet angle of the air inlet pipe. The present application can collect a variety of air environment data in real time and make dynamic adjustments to ensure that the best air sample can be captured under any circumstances. At the same time, through the combination of intelligent algorithms and PLC controllers, the system can quickly respond to environmental changes and improve monitoring efficiency and accuracy.

[0019] Figure 1 FIG. 1 is a block diagram of a controller in a real-time online monitoring system for building carbon emissions according to an embodiment of the present application. Figure 1 As shown, the controller 100 includes: an air environment data receiving module 110, used to receive the air environment data collected by the sensor; an air environment data timing collaborative analysis module 120, used to perform a core timing collaborative analysis of the air environment data based on the carbon emission main variable-environmental data auxiliary variable to obtain a recommended air outlet angle value; an air inlet duct air outlet angle adjustment module 130, used to input the recommended air outlet angle value and the real-time air outlet angle of the air inlet duct into a PLC controller to obtain a control instruction, and the control instruction is used to adjust the air outlet angle of the air inlet duct.

[0020] In the embodiment of the present application, the air environment data receiving module 110 is used to receive the air environment data collected by the sensor. In particular, the air environment data includes a time series of wind speed values, a time series of wind direction values, a time series of ambient temperature values, a time series of ambient humidity values, and a time series of carbon emission concentration values. It should be understood that the time series of wind speed values ​​can reflect the speed and change of air flow. Different wind speeds will affect the mixing degree of air and the diffusion range of pollutants. For example, when the wind speed is high, the air flows quickly, and carbon emissions may be quickly diluted and diffused to a larger area; while when the wind speed is low, carbon emissions may accumulate in a local area. The time series of wind direction values ​​can clarify the direction and change trend of air. The air in different directions around the building may have different carbon emission characteristics, which is related to factors such as the distribution of surrounding pollution sources and topography. For example, if there is a factory on one side of the building, when the wind direction points to that side, the carbon emission concentration of the air collected by the air inlet pipe may be higher. The time series of ambient temperature values ​​reflects the changes in the thermal environment. Temperature affects the density and vertical movement of air, thereby affecting the vertical distribution of carbon emissions. For example, during the day when the temperature is high, the air on the ground rises due to the heat, which may drive the carbon emissions at the bottom to diffuse upward. The time series of ambient humidity values ​​helps to understand the changes in the water vapor content in the air. Humidity affects the physical and chemical properties of certain pollutants and is also related to the distribution of carbon emissions. For example, in a high humidity environment, some pollutants may attach to water vapor particles, changing their transmission and distribution in the air. The time series of carbon emission concentration values ​​directly reflects the actual situation and changing trend of carbon emissions around the building, and is the core indicator of the monitoring system. In general, by obtaining and analyzing these air environment data, the system can perceive the changes in the environment in real time, and based on this, more accurately adjust the angle of the air inlet pipe to make the collected air samples more representative.

[0021] In the embodiment of the present application, the air environment data time series collaborative analysis module 120 is used to perform a core time series collaborative analysis based on the carbon emission main variable and the environmental data auxiliary variable on the air environment data to obtain a recommended air outlet angle value. Specifically, Figure 2 It is a block diagram of the air environment data time series collaborative analysis module in the real-time online monitoring system of building carbon emissions according to an embodiment of the present application. Figure 3 This is a data flow diagram of the air environment data time series collaborative analysis module in the real-time online monitoring system for building carbon emissions according to an embodiment of the present application. Figure 2 and Figure 3As shown, the air environment data time series collaborative analysis module 120 includes: an air environment data grouping unit 121, which is used to perform variable grouping on the air environment data to obtain a time series of carbon emission concentration values ​​and a time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value}; a carbon emission auxiliary variable encoding unit 122, which is used to perform carbon emission auxiliary variable time series feature encoding on the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} to obtain a time series collaborative feature vector between carbon emission auxiliary variables; a carbon emission concentration time series encoding unit 123, Used to extract the carbon emission concentration time series correlation characteristics from the time series of the carbon emission concentration value to obtain the carbon emission concentration time series correlation characteristic coding vector; the time series depth coordination unit 124 is used to perform carbon emission main-auxiliary variable core time series depth coordination analysis on the carbon emission auxiliary variable time series coordination feature vector and the carbon emission concentration time series correlation characteristic coding vector to obtain the carbon emission main-auxiliary variable time series depth coordination coding feature vector; the air outlet angle value recommendation unit 125 is used to obtain the recommended air outlet angle value based on the carbon emission main-auxiliary variable time series depth coordination coding feature vector.

[0022] In an embodiment of the present application, the air environment data grouping unit 121 is used to group the air environment data by variables to obtain a time series of carbon emission concentration values ​​and a time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value}. Accordingly, it is considered that there are obvious differences between the carbon emission concentration value and the wind speed value, wind direction value, ambient temperature value, and ambient humidity value in data characteristics and the physical meaning represented. Specifically, the carbon emission concentration value directly reflects the content of carbon-related substances in the air and is a key indicator for measuring carbon emissions; while the wind speed value reflects the air flow speed, the wind direction value indicates the air flow direction, the ambient temperature value reflects the degree of coldness and heat of the air, and the ambient humidity value represents the water vapor content in the air. They are environmental factors that affect the carbon emission concentration and belong to auxiliary variables. Based on this, in order to be able to more clearly sort out the relationship between the data, the present application obtains a time series of carbon emission concentration values ​​and a time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} by grouping the air environment data by variables. In this way, the relationship between carbon emission concentration values ​​and other environmental factors can be studied more specifically. For example, by analyzing the changing patterns of carbon emission concentration values ​​under different wind speeds, wind directions, temperatures and humidity conditions, we can identify the key environmental factors affecting carbon emission concentrations and their mechanisms of action, thereby providing more in-depth theoretical support for the monitoring and control of carbon emissions.

[0023] In an embodiment of the present application, the carbon emission auxiliary variable encoding unit 122 is used to perform carbon emission auxiliary variable time series feature encoding on the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} to obtain a time series collaborative feature vector between carbon emission auxiliary variables. Specifically, in an embodiment of the present application, the carbon emission auxiliary variable encoding unit is used to: input the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} into a carbon emission auxiliary variable time series feature encoder based on a convolutional neural network model to obtain a time series collaborative feature vector between carbon emission auxiliary variables. Accordingly, considering that the time series of variables such as wind speed, wind direction, ambient temperature and ambient humidity have certain local correlations in the time dimension, for example, the change of wind speed may be continuous in a short time, and the fluctuation of temperature also has certain local characteristics. Moreover, these four variables are not independent of each other, but have a complex collaborative relationship. For example, wind speed and wind direction will affect the flow of air, and then affect the distribution of ambient temperature and humidity; changes in temperature and humidity may also affect the density and flow of air, thereby indirectly affecting wind speed and wind direction. Based on this, in the technical solution of the present application, the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} is input into the carbon emission auxiliary variable time series feature encoder based on the convolutional neural network model to obtain the time series collaborative feature vector between carbon emission auxiliary variables. It should be understood that the convolutional neural network has unique advantages in processing time series data with local correlation. Its convolution layer can slide on the time series through the convolution kernel to automatically capture these local feature patterns. At the same time, it can automatically learn and capture the collaborative change patterns between these variables in the time series through its multi-layer structure and convolution operation, and dig out the complex relationship hidden behind the data. In this way, the obtained time series collaborative feature vector between carbon emission auxiliary variables can more comprehensively and accurately reflect the collaborative changes of environmental factors such as wind speed, wind direction, temperature and humidity in the time series, and help the model better understand the impact mechanism of environmental factors on carbon emissions, thereby improving the accuracy and reliability of carbon emission analysis.

[0024] The following is a detailed description of a specific implementation process of "inputting the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} into a carbon emission auxiliary variable time series feature encoder based on a convolutional neural network model to obtain a time series collaborative feature vector between the carbon emission auxiliary variables": The first is the data preprocessing stage. The main task of this stage is to standardize or normalize the raw data so that different types of environmental parameters can be compared and analyzed on the same scale. In addition, since data collected over a long time span will inevitably have missing values, effective measures need to be taken to fill these gaps. Common methods include linear interpolation or more complex machine learning prediction methods, and the most suitable technical means are selected according to the specific situation. The purpose of this is to ensure the integrity of the data set input into the subsequent model and avoid analysis bias caused by incomplete data.

[0025] Once the data preprocessing is completed, the process of constructing the input tensor begins. The focus here is on how to organize the cleaned and formatted time series of wind speed values, wind direction values, ambient temperature values, and ambient humidity values ​​into a form suitable for CNN model processing. Usually, each type of environmental data is input into the CNN model as a separate channel. This means that for each time point, there is a vector containing four values, representing the wind speed, wind direction, temperature, and humidity at that moment. This multi-channel design helps the model better understand the interaction between various environmental factors and their impact on the ultimate goal, that is, carbon emission concentration.

[0026] Next, it is necessary to design an efficient convolutional neural network structure. This process involves several key links: first, the design of the convolutional layer, where the number and size of filters used in each layer need to be determined. These filters will slide on the time dimension of the input data to capture local patterns and features in the time series. It is worth noting that it is very important to apply nonlinear activation functions such as ReLU after each convolutional layer, because this can increase the learning ability and expression ability of the model. Subsequently, adding pooling layers after some convolutional layers, such as the maximum pooling layer, can help reduce the spatial dimension of the feature map, while reducing the complexity of the model and preventing overfitting. Finally, in order to map the previously extracted features to the final output, one or more fully connected layers are added at the end of the network, which are responsible for integrating all information and generating the final time series collaborative feature vector between carbon emission auxiliary variables.

[0027] When the entire network architecture is built, it is time to train the model. In this process, the labeled historical data set will be used as training samples, and the network weights will be continuously adjusted through the back propagation algorithm so that the model can accurately learn relevant features from the input time series that are helpful for predicting carbon emissions. This is an iterative optimization process, and it may take multiple experiments to find the optimal hyperparameter configuration. It is worth noting that during training, a validation set needs to be set to monitor the performance of the model to avoid overfitting problems.

[0028] After the model training is completed, it can be applied to the newly collected environmental data to generate a time series collaborative feature vector between carbon emission auxiliary variables. Specifically, when there is a time series of new wind speed values, wind direction values, ambient temperature values, and ambient humidity values, just input these data into the model in the previously set way, and the last few layers of the model will output a set of values, which is the required feature vector. It reflects the relevant time series characteristics of carbon emission auxiliary variables in the original time series, which can provide strong support for subsequent analysis.

[0029] In an embodiment of the present application, the carbon emission concentration time series encoding unit 123 is used to extract the carbon emission concentration time series correlation characteristics from the time series of the carbon emission concentration value to obtain the carbon emission concentration time series correlation characteristic encoding vector. Specifically, in an embodiment of the present application, the carbon emission concentration time series encoding unit is used to: use a TCN-based time series encoder to extract the carbon emission concentration time series correlation characteristics from the time series of the carbon emission concentration value to obtain the carbon emission concentration time series correlation characteristic encoding vector. Accordingly, considering that the time series of the carbon emission concentration value itself has certain dynamic change characteristics, there is a temporal correlation relationship between the concentration values ​​at adjacent time points. For example, the carbon emission concentration at a certain moment may be affected by the concentration values ​​at previous moments, or show certain periodic and trend changes. These changes may be affected by many factors, such as activities in the building, the operating status of the ventilation system, and external meteorological conditions. Therefore, in order to capture the mutual relationship and pattern between carbon emission concentration values ​​at different time points, so as to better understand its dynamic change law, the present application extracts the carbon emission concentration time series correlation features from the time series of the carbon emission concentration values ​​to deeply explore the inherent laws and characteristics hidden in the data, and obtains the carbon emission concentration time series correlation feature encoding vector. In particular, in a specific example of the present application, the carbon emission concentration time series correlation features can be extracted from the time series of the carbon emission concentration values ​​by using a TCN-based time encoder to obtain the carbon emission concentration time series correlation feature encoding vector. It can be understood that in the time series data of carbon emission concentration values, the carbon emission concentration at the current moment is often associated with the concentration value over a long period of time in the past. TCN has a strong ability to capture long-term time dependencies. By stacking multiple convolutional layers and dilated convolution operations, it can effectively model the information in a longer time series, and can mine the dependency of carbon emission concentrations on different time scales, so as to better characterize the changing trends and laws of carbon emission concentrations.

[0030] In the embodiment of the present application, the time series depth coordination unit 124 is used to perform a carbon emission main-auxiliary variable core time series depth coordination analysis on the carbon emission auxiliary variable time series coordination feature vector and the carbon emission concentration time series correlation feature coding vector to obtain a carbon emission main-auxiliary variable time series depth coordination coding feature vector. Specifically, Figure 4 FIG. 1 is a block diagram of a time series depth coordination unit in a real-time online monitoring system for building carbon emissions according to an embodiment of the present application. Figure 4 As shown, the time series deep coordination unit 124 includes: a carbon emission auxiliary-main variable time series multi-granularity interaction subunit 1241, which is used to construct the characteristic granularity response interaction characteristics and characteristic value granularity response interaction characteristics between the carbon emission auxiliary variable time series coordination feature vector and the carbon emission concentration time series correlation feature coding vector to obtain the carbon emission auxiliary-main variable time series characteristic granularity response interaction coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interaction coding vector; a carbon emission auxiliary-main variable time series deep coordination coding subunit 1242, which is used to fuse the carbon emission auxiliary-main variable time series characteristic granularity response interaction coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interaction coding vector to obtain the carbon emission main-auxiliary variable time series deep coordination coding feature vector.

[0031] It should be understood that carbon emission concentration is the main variable, and auxiliary variables such as wind speed, wind direction, ambient temperature and humidity influence and interact with each other in the actual environment. Analyzing the main variable or auxiliary variable alone can only obtain partial information, and there is a complex time series synergy relationship between them. For example, wind speed and wind direction will affect the diffusion of carbon emission substances, and then affect the carbon emission concentration at different times; ambient temperature and humidity may affect the rate of chemical reactions related to carbon emissions and change the trend of carbon emission concentration. At the same time, although the characteristic vectors of the main and auxiliary variables of carbon emissions contain certain information, there may be hidden and underutilized correlation information. Based on this, in order to discover the deep interaction patterns and laws between variables, so as to capture the real correlation between these variables in the time dimension more comprehensively and accurately. The present application performs a core time series deep synergy analysis of the carbon emission main-auxiliary variables on the time series synergy feature vector between the carbon emission auxiliary variables and the time series correlation feature coding vector of the carbon emission concentration to obtain the time series deep synergy coding feature vector of the carbon emission main-auxiliary variables. In particular, when this method performs deep collaborative analysis of the core time series of two types of feature vectors, it first uses autocorrelation decoupling technology to extract core anchor points such as wind speed mutations and concentration peaks. By using two-level interactive modeling, the correlation between the main and auxiliary variable features is quantified. After feature fusion, a highly compressed and structured coding vector is generated, which can accurately understand the laws of carbon emissions and help make efficient decisions and precise regulation.

[0032] Specifically, in the embodiment of the present application, the carbon emission auxiliary-main variable time series multi-granularity interaction subunit is used to: construct the semantic autocorrelation association matrix of the carbon emission auxiliary variable time series collaborative feature vector and the carbon emission concentration time series associated feature coding vector to obtain the carbon emission auxiliary variable time series collaborative semantic autocorrelation association matrix and the carbon emission concentration time series associated semantic autocorrelation association matrix. The process can be expressed by the formula: in, is the time series collaborative feature vector between the auxiliary variables of carbon emission, is the carbon emission concentration time series correlation feature coding vector, is the transpose operation, and is a feature mapping function, such as a linear mapping or a nonlinear kernel function, is the temporal collaborative semantic autocorrelation matrix of carbon emission auxiliary variables, is the semantic autocorrelation matrix of carbon emission concentration temporal association; The carbon emission auxiliary variable time series collaborative semantic autocorrelation association matrix and the carbon emission concentration time series correlation semantic autocorrelation association matrix are respectively subjected to core time series autocorrelation decoupling to obtain the carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector. The process can be expressed by the formula: in, is the temporal collaborative semantic autocorrelation matrix of carbon emission auxiliary variables, is the carbon emission concentration temporal correlation semantic autocorrelation matrix, To perform core timing autocorrelation decoupling operations, For feature decoupling operations, , , and They are respectively Each row vector of and They are The corresponding carbon emission auxiliary variable weight matrix and carbon emission auxiliary variable bias vector, is matrix multiplication, is the carbon emission auxiliary variable scoring weight vector, It is the first in the set of time-series significant factors among auxiliary variables of carbon emissions. The significant time series factors among the auxiliary variables of carbon emissions, is the normalization function, It is the first in the set of time-series significant factors between auxiliary variables of normalized carbon emissions. Normalized carbon emission auxiliary variables have significant time series factors, is the The number of row vectors in is the core time-series anchor coding vector of carbon emission auxiliary variables, , , and They are respectively Each row vector of and They are The corresponding carbon emission concentration weight matrix and carbon emission concentration bias vector, is the carbon emission concentration score weight vector, It is the first in the set of significant factors of carbon emission concentration. Significant factors of carbon emission concentration, It is the first in the set of significant factors of normalized carbon emission concentration. The significant factor of normalized carbon emission concentration is is the The number of row vectors in , and and The number of is the core temporal anchor coding vector of carbon emission concentration; The carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector are subjected to characteristic granularity response interactive coding to obtain the carbon emission auxiliary-primary variable time series characteristic granularity response interactive coding vector. The process can be expressed by the formula: in, is the core time-series anchor coding vector of carbon emission auxiliary variables, is the core temporal anchor coding vector of carbon emission concentration, It is added by location point. and are the response interaction weight matrix and the response interaction bias vector, respectively. is the hyperbolic tangent function, is the carbon emission auxiliary-primary variable time series characteristic granularity response interaction coding vector; The carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector are interactively coded with characteristic value granularity response to obtain the carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector. The process can be expressed by the formula: in, is the core time series anchor coding vector of carbon emission auxiliary variables, is the core temporal anchor coding vector of carbon emission concentration, It is the interaction coding vector of the granular response of the time series characteristic values ​​of the main and auxiliary variables of carbon emissions.

[0033] It should be understood that the original carbon emission auxiliary variable time series synergy feature vector and carbon emission concentration time series correlation feature coding vector have relatively obscure internal connections. By constructing their own semantic autocorrelation association matrices, the mutual connections between the components in the original feature vector can be transformed from implicit to explicit, and clearly displayed in the form of a matrix. This method enables the model to more intuitively understand the connection between the various parts of the vector, and can provide more comprehensive information for subsequent in-depth analysis, avoiding analysis bias caused by ignoring the connection between the elements within the vector.

[0034] Accordingly, considering that the generated carbon emission auxiliary variable time series collaborative semantic autocorrelation matrix and carbon emission concentration time series correlation semantic autocorrelation matrix contain both key parts closely related to the core semantics and redundant information that may interfere with the analysis. In order to be able to screen and refine these complex information, it is necessary to perform core time series autocorrelation decoupling processing on these two semantic autocorrelation matrices in this application. Specifically, it can highlight the parts that are highly related to the core semantics with the help of feature distillation, and perform noise reduction processing on redundant information, so as to extract more valuable and structurally refined core anchor point representations from complex matrix information. That is, the "anchor points" in the two encoding vectors of the carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector can effectively capture the global information of the carbon emission auxiliary variable time series features and the carbon emission concentration time series features, and ensure the semantic consistency of feature representation in the compressed space. In essence, by constraining the information entropy, the uncertainty of the information is reduced, so that the model can be applied more stably and accurately in different carbon emission monitoring scenarios, and the generalization ability of the model is improved, which can provide more high-quality and concise basic data for subsequent feature interaction analysis.

[0035] It should be understood that after completing the anchoring of the core information, in order to gain a deeper understanding of the relationship between the auxiliary variables of carbon emissions and the carbon emission concentration, it is necessary to interactively analyze their core temporal anchor encoding vectors. At the feature granularity level, the feature vector can be regarded as composed of multiple semantic units. By exploring the response relationship between the two feature vectors on these semantic units through interactive encoding, the semantic coupling between them can be accurately captured from a micro level. This analysis method is similar to the multimodal embedding task, but it focuses more on the interaction of homogeneous features. It is not just a simple point-to-point comparison of a single semantic unit, but also comprehensively considers its overall behavior pattern in the entire feature set, so as to understand the semantic connection between features more comprehensively and deeply. That is, the generated carbon emission auxiliary-main variable temporal feature granularity response interactive encoding vector reflects in detail the interaction of the two feature vectors at the micro semantic level, clearly shows the response relationship between different semantic units, and can provide data support from a micro perspective for a deep understanding of the semantic association between the main and auxiliary variables of carbon emissions, which is conducive to enriching the model's understanding of the relationship between the two.

[0036] Accordingly, considering that in building carbon emission monitoring, the relationship between carbon emission auxiliary variables (wind speed, wind direction, temperature, humidity, etc.) and carbon emission concentration is extremely complex. Relying solely on high-level semantic analysis cannot fully reveal the subtle correlations between the eigenvalues ​​of these variables. In order to be able to deeply explore these interdependencies hidden in the eigenvalue data, this application needs to perform eigenvalue granularity response interactive coding operations on the core time series anchor coding vector of the carbon emission auxiliary variable and the core time series anchor coding vector of the carbon emission concentration. Specifically, through the element-by-element division operation, the relationship between the eigenvalues ​​of the two vectors is carefully characterized to further enrich the understanding of the relationship between the main and auxiliary variables of carbon emissions. That is, the generated carbon emission main-auxiliary variable time series eigenvalue granularity response interactive coding vector reveals the interdependence between the eigenvectors from a numerical level, makes up for the lack of analysis of the eigenvalue granularity response interactive coding vector at the numerical level, and further enriches the representation dimension of the interactive modeling, which can provide more comprehensive data support for a comprehensive understanding of the complex relationship between the main and auxiliary variables of carbon emissions.

[0037] Finally, the carbon emission auxiliary-main variable time series feature granularity response interactive coding vector and the carbon emission main-auxiliary variable time series feature value granularity response interactive coding vector are fused to obtain the carbon emission main-auxiliary variable time series deep collaborative coding feature vector. The above process can be expressed as: in, is the carbon emission auxiliary-primary variable time series characteristic granularity response interaction coding vector, is the interaction coding vector of the granular response of the time series characteristic value of the main and auxiliary variables of carbon emissions, It is the time series deep collaborative coding feature vector of the main and auxiliary variables of carbon emissions.

[0038] It should be understood that after the interactive encoding of the characteristic granularity and the characteristic value granularity response, although the interactive information of the main and auxiliary variables of carbon emissions is obtained from the semantic and numerical levels, this information is scattered in different coding vectors. In order to fully and comprehensively reflect the relationship between variables, it is necessary to integrate these multi-granular information. By cascading the carbon emission auxiliary-main variable time series characteristic granularity response interactive coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector, a unified coding vector can be constructed to have the ability to express multi-granular information. In other words, the generated carbon emission main-auxiliary variable time series deep collaborative coding feature vector integrates multi-granular information at the semantic and numerical levels, and can more comprehensively and accurately reflect the complex time series deep collaborative relationship between the main and auxiliary variables of carbon emissions, which can provide strong data support for accurate insight into the laws of carbon emissions.

[0039] Preferably, in another example of the present application, the carbon emission auxiliary-main variable time series deep collaborative coding subunit is used to: perform gradient contribution disturbance correction on the carbon emission auxiliary-main variable time series characteristic granularity response interactive coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector to obtain the corrected carbon emission auxiliary-main variable time series characteristic granularity response interactive coding vector and the corrected carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector; fuse the corrected carbon emission auxiliary-main variable time series characteristic granularity response interactive coding vector and the corrected carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector to obtain the carbon emission main-auxiliary variable time series deep collaborative coding feature vector. The processing process of the carbon emission auxiliary-main variable time series deep collaborative coding subunit is expressed by the formula: in, yes Middle The eigenvalues ​​at the positions, yes Middle The eigenvalues ​​at the positions, is the cosine function, yes The corrected eigenvalues, yes The corrected eigenvalues, is the interaction coding vector of the time series characteristic granularity response of auxiliary and main variables of carbon emissions after correction, is the granular response interaction coding vector of the time series characteristic value of the main and auxiliary variables of carbon emissions after correction, It is the time series deep collaborative coding feature vector of the main and auxiliary variables of carbon emissions.

[0040] In particular, here, considering the difference between the modeling representation of the characteristic granularity interaction of carbon emission variables and the eigenvalue granularity interaction, it may cause unstable perturbations in the characteristic manifold interface of the fused carbon emission main-auxiliary variable time series deep collaborative coding feature vector with multi-granularity information expression. The eigenvalue granularity interaction representation of carbon emission variables is used as an extensibility constraint representation to grow the overall characteristic interaction distribution under the eigenvalue diffusion process. Based on the growth exponential representation under the extensibility constraint, the interface shape perturbation deviation of the overall distribution of the characteristic granularity interaction of carbon emission variables is modeled. That is, the interaction interface gradient under the eigenvalue granularity is used as the perturbation contribution factor, and the growth exponential stabilization contribution of the gradient diffusion under the eigenvalue mutual dependence is determined to achieve the growth mode gradient correction under the dominance of perturbation stabilization, thereby improving the manifold interface stability of the carbon emission main-auxiliary variable time series deep collaborative coding feature vector.

[0041] In an embodiment of the present application, the tuyere angle value recommendation unit 125 is used to obtain the recommended tuyere angle value based on the carbon emission main-auxiliary variable time series deep collaborative coding feature vector. Specifically, in an embodiment of the present application, the tuyere angle value recommendation unit is used to: input the carbon emission main-auxiliary variable time series deep collaborative coding feature vector into the tuyere angle optimizer based on the decoder to obtain the recommended tuyere angle value. That is to say, the carbon emission main-auxiliary variable time series deep collaborative coding feature vector obtained by core time series collaborative analysis using the carbon emission auxiliary variable time series collaborative feature vector and the carbon emission concentration time series associated feature coding vector is decoded to intelligently obtain the recommended tuyere angle value. It should be understood that the decoder-based structure has advantages in processing tasks that require generating specific outputs from complex feature representations. For the task of tuyere angle optimization, there is a complex nonlinear relationship between its output (tuyere angle value) and many environmental factors and carbon emission concentrations. By learning this nonlinear mapping relationship, the decoder can accurately calculate the tuyere angle value that is compatible with the current environmental state and carbon emission situation based on the input feature vector. In a specific embodiment of the present application, the carbon emission main-auxiliary variable time series deep collaborative coding feature vector is input into a decoder-based tuyere angle optimizer to obtain the recommended tuyere angle value, including: multiplying the decoding weight matrix of the decoder with the carbon emission main-auxiliary variable time series deep collaborative coding feature vector to obtain the carbon emission main-auxiliary variable time series deep collaborative decoding feature vector, and accumulating and summing all eigenvalues ​​of the carbon emission main-auxiliary variable time series deep collaborative decoding feature vector to obtain the recommended tuyere angle value.

[0042] In the embodiment of the present application, the air inlet angle adjustment module 130 is used to input the recommended air outlet angle value and the real-time air outlet angle of the air inlet pipe into the PLC controller to obtain a control instruction, and the control instruction is used to adjust the air outlet angle of the air inlet pipe. It should be understood that the PLC (programmable logic controller) controller has powerful logic control and precise operation capabilities. The recommended air outlet angle value is an ideal angle obtained based on a deep analysis of the main and auxiliary variables of carbon emissions, and the real-time air outlet angle of the air inlet pipe reflects the current actual state. Inputting these two key information into the PLC controller enables it to accurately calculate the control parameters required to adjust the air inlet angle of the air inlet pipe according to the preset logic and algorithm. Compared with other control methods, the PLC controller can perform the angle adjustment task more accurately and stably, ensuring that the air outlet angle of the air inlet pipe can be accurately adjusted as expected. In this way, adjusting the air outlet angle of the air inlet pipe through the control instruction generated by the PLC controller can enable the air inlet pipe to better adapt to environmental changes and optimize the air collection effect. For example, when the wind direction changes, timely adjustment of the air outlet angle can ensure that the air inlet duct can collect representative air samples, avoiding the situation where the collected air cannot accurately reflect the actual carbon emissions due to improper air outlet angle, thereby improving the accuracy and reliability of carbon emission monitoring data.

[0043] The following is a detailed description of a specific implementation process of "inputting the recommended air outlet angle value and the real-time air outlet angle of the air inlet duct into a PLC controller to obtain a control instruction, wherein the control instruction is used to adjust the air outlet angle of the air inlet duct": First, the hardware connection work must be completed. This requires connecting the angle sensor that can accurately measure the real-time air outlet angle of the air inlet duct to the PLC controller. The angle sensor will transmit the current angle information of the air inlet duct to the PLC controller in the form of an electrical signal or a digital signal, so that the controller knows the real-time status of the air inlet duct. At the same time, the communication interface for receiving the recommended air outlet angle value must also be connected to the PLC controller. The recommended air outlet angle value is obtained by other modules of the system after in-depth analysis and processing of the air environment data, and will be sent through network communication or other data transmission methods.

[0044] After the data is successfully transmitted to the PLC controller, it will be stored in its internal specific register or data storage area. The PLC controller will then process the data according to the control logic and algorithm set in advance. Specifically, the difference between the recommended air outlet angle value and the real-time air outlet angle will be calculated. This difference is crucial, as it determines the direction and magnitude of the air inlet duct adjustment. For example, if the recommended angle is 30 degrees and the real-time angle is 20 degrees, it means that the air inlet duct needs to be adjusted 10 degrees in the direction of increasing the angle.

[0045] Based on the calculated angle difference, the PLC controller will generate control instructions according to the pre-programmed program logic. This control instruction contains very precise control parameters, such as the motor's rotation direction, rotation speed, and rotation time. These parameters directly determine the specific actions of the air inlet duct adjustment. If it is calculated that the air inlet duct needs to rotate 5 degrees counterclockwise, the control instruction will make the drive motor rotate counterclockwise by the corresponding angle, thereby driving the air inlet duct to the appropriate position.

[0046] The generated control command will be output to the drive actuator connected to the air inlet duct. The most common drive actuator is the motor. After receiving the control command, the motor will start to operate according to the parameters in the command. The motor converts its own rotational motion into the angle adjustment motion of the air inlet duct. In this process, the operation accuracy and stability of the motor play a key role, which is directly related to the accuracy of the angle adjustment of the air inlet duct. If the motor runs unstably, the angle adjustment of the air inlet duct may deviate, affecting the monitoring effect.

[0047] During the process of adjusting the air outlet angle of the air inlet duct, the angle sensor will continuously monitor the real-time angle of the air inlet duct and feed back the new angle data to the PLC controller. The PLC controller will compare the current angle and the recommended angle in real time. Once it finds that the difference between the two is within the allowable error range, it will stop sending control instructions to stabilize the air inlet duct at the recommended air outlet angle position. However, if the deviation is too large or an abnormal situation occurs during the adjustment process, the PLC controller will make corresponding adjustments or issue an alarm message according to the preset exception handling procedure to ensure the reliability and safety of the entire adjustment process. For example, if the angle feedback from the angle sensor deviates from the recommended angle by more than 5 degrees, it is considered a large deviation. The PLC controller may recalculate the control instruction and increase the rotation amplitude of the motor to adjust the air inlet duct to the appropriate angle as soon as possible.

[0048] The entire air inlet angle adjustment process is continuous. The sensor will continuously collect new air environment data, and after analysis and processing by the controller, a new recommended air outlet angle value will be generated. The PLC controller will continue to receive the recommended air outlet angle value and the real-time air inlet angle data of the air inlet, and repeat the above adjustment process. In this way, the air inlet angle of the air inlet can be adjusted dynamically in real time according to environmental changes, ensuring that the monitoring system can always collect representative air samples, thereby improving the accuracy and reliability of carbon emission monitoring data.

[0049] In summary, the controller 100 is clearly explained, and it uses data analysis and encoding based on deep learning to group air environment data into variables to obtain the time series of carbon emission concentration values ​​and the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value}, and then, the two time series are time-series encoded, so as to intelligently obtain the recommended air outlet angle value based on the time series synergy characteristics between the encoded carbon emission auxiliary variables and the core time series deep synergy representation between the time series correlation characteristics of the carbon emission concentration, and input it into the PLC controller with the real-time air outlet angle to adaptively adjust the air outlet angle of the air inlet duct. In this way, a variety of air environment data can be collected in real time and dynamically adjusted to ensure that the best air sample can be captured under any circumstances. At the same time, through the combination of intelligent algorithms and PLC controllers, the system can respond quickly to environmental changes and improve monitoring efficiency and accuracy.

[0050] In summary, a real-time online monitoring system for building carbon emissions based on the embodiment of the present application is explained, which includes a detection box, a controller, a negative pressure fan and a sensor installed inside the detection box, an air inlet is provided on the top wall of the detection box, an air inlet pipe is installed inside, a filter is provided at the air inlet end of the air inlet pipe, and exhaust pipe ports are provided on both sides of the detection box, wherein the sensor is used to collect air environment data, and the controller is used to adjust the air outlet angle of the air inlet pipe by analyzing the collected air environment data. In this way, by dynamically adjusting the air outlet angle of the air inlet pipe, rapid response to environmental changes and more accurate monitoring data collection can be achieved.

Claims

1. A real-time online monitoring system for building carbon emissions, comprising a detection box, a negative pressure fan and a sensor installed inside the detection box, wherein: The top wall of the detection box is provided with an air inlet, an air inlet pipe is provided in the air inlet, a filter is installed at the air inlet end of the air inlet pipe, and exhaust pipe openings are provided on both side walls of the detection box. The sensor is used to collect air environment data. It is characterized in that the real-time online monitoring system for building carbon emissions also includes a controller, and the controller is used to adjust the air outlet angle of the air inlet pipe; Wherein, the controller comprises: An air environment data receiving module, used to receive the air environment data collected by the sensor; An air environment data time series collaborative analysis module is used to perform a core time series collaborative analysis on the air environment data based on the carbon emission main variable and the environmental data auxiliary variable to obtain a recommended air outlet angle value; The air inlet angle adjustment module is used to input the recommended air inlet angle value and the real-time air inlet angle of the air inlet duct into the PLC controller to obtain a control instruction, and the control instruction is used to adjust the air inlet angle of the air inlet duct.

2. The real-time online monitoring system for building carbon emissions according to claim 1 is characterized in that: The air environment data includes a time series of wind speed values, a time series of wind direction values, a time series of ambient temperature values, a time series of ambient humidity values, and a time series of carbon emission concentration values.

3. The real-time online monitoring system for building carbon emissions according to claim 2 is characterized in that: The air environment data time series collaborative analysis module includes: An air environment data grouping unit, used for performing variable grouping on the air environment data to obtain a time series of carbon emission concentration values ​​and a time series of {wind speed values, wind direction values, ambient temperature values, ambient humidity values}; A carbon emission auxiliary variable encoding unit, used for performing carbon emission auxiliary variable time series feature encoding on the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} to obtain a time series collaborative feature vector between carbon emission auxiliary variables; A carbon emission concentration time series encoding unit, used to extract carbon emission concentration time series correlation features from the time series of the carbon emission concentration values ​​to obtain a carbon emission concentration time series correlation feature encoding vector; A time series deep coordination unit, used for performing a carbon emission main-auxiliary variable core time series deep coordination analysis on the carbon emission auxiliary variable time series coordination feature vector and the carbon emission concentration time series correlation feature coding vector to obtain a carbon emission main-auxiliary variable time series deep coordination coding feature vector; The air vent angle value recommendation unit is used to obtain the recommended air vent angle value based on the carbon emission main-auxiliary variable time series deep collaborative coding feature vector.

4. The real-time online monitoring system for building carbon emissions according to claim 3 is characterized in that: The carbon emission auxiliary variable encoding unit is used to: input the time series of {wind speed value, wind direction value, ambient temperature value, ambient humidity value} into a carbon emission auxiliary variable time series feature encoder based on a convolutional neural network model to obtain a time series collaborative feature vector between the carbon emission auxiliary variables.

5. The real-time online monitoring system for building carbon emissions according to claim 4 is characterized in that: The carbon emission concentration time series encoding unit is used to: use a TCN-based time series encoder to extract carbon emission concentration time series correlation features from the time series of the carbon emission concentration values ​​to obtain the carbon emission concentration time series correlation feature encoding vector.

6. The real-time online monitoring system for building carbon emissions according to claim 5 is characterized in that: The time series depth coordination unit includes: The carbon emission auxiliary-main variable time series multi-granularity interaction subunit is used to construct the characteristic granularity response interaction characteristics and characteristic value granularity response interaction characteristics between the carbon emission auxiliary variable time series collaborative characteristic vector and the carbon emission concentration time series associated characteristic coding vector to obtain the carbon emission auxiliary-main variable time series characteristic granularity response interaction coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interaction coding vector; The carbon emission auxiliary-main variable time series deep collaborative coding subunit is used to fuse the carbon emission auxiliary-main variable time series feature granularity response interactive coding vector and the carbon emission main-auxiliary variable time series feature value granularity response interactive coding vector to obtain the carbon emission main-auxiliary variable time series deep collaborative coding feature vector.

7. The real-time online monitoring system for building carbon emissions according to claim 6 is characterized in that: The carbon emission auxiliary-primary variable time series multi-granularity interaction sub-unit is used to: Constructing a semantic autocorrelation association matrix of the time series synergy feature vector between the carbon emission auxiliary variables and the time series association feature coding vector of the carbon emission concentration to obtain a carbon emission auxiliary variable time series synergy semantic autocorrelation association matrix and a carbon emission concentration time series association semantic autocorrelation association matrix; Performing core time series autocorrelation decoupling on the carbon emission auxiliary variable time series collaborative semantic autocorrelation association matrix and the carbon emission concentration time series correlation semantic autocorrelation association matrix respectively to obtain a carbon emission auxiliary variable core time series anchor coding vector and a carbon emission concentration core time series anchor coding vector; Performing feature granularity response interactive coding on the carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector to obtain the carbon emission auxiliary-primary variable time series feature granularity response interactive coding vector; The carbon emission auxiliary variable core time series anchor coding vector and the carbon emission concentration core time series anchor coding vector are interactively coded for characteristic value granularity response to obtain the carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector.

8. The real-time online monitoring system for building carbon emissions according to claim 7 is characterized in that: The carbon emission auxiliary-primary variable time series deep collaborative coding subunit is used to: Performing gradient contribution disturbance correction on the carbon emission auxiliary-primary variable time series characteristic granularity response interaction coding vector and the carbon emission main-auxiliary variable time series characteristic value granularity response interaction coding vector to obtain a corrected carbon emission auxiliary-primary variable time series characteristic granularity response interaction coding vector and a corrected carbon emission main-auxiliary variable time series characteristic value granularity response interaction coding vector; The corrected carbon emission auxiliary-main variable time series characteristic granularity response interactive coding vector and the corrected carbon emission main-auxiliary variable time series characteristic value granularity response interactive coding vector are fused to obtain the carbon emission main-auxiliary variable time series deep collaborative coding feature vector.

9. The real-time online monitoring system for building carbon emissions according to claim 8 is characterized in that: The tuyere angle value recommendation unit is used to: input the carbon emission main-auxiliary variable time series deep collaborative coding feature vector into the decoder-based tuyere angle optimizer to obtain the recommended tuyere angle value.

Images