Airport terminal carbon emission prediction method and system

Abstract

The invention relates to the technical field of building carbon emission prediction, and provides an airport terminal carbon emission prediction method and system, and the method comprises the steps: obtaining the key data of an airport terminal at the current moment; inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain predicted carbon emission at the current moment; wherein the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emission at the previous moment and the actually measured carbon emission at the previous moment; and a Markov model in the LSTM-Markov carbon emission prediction model dynamically corrects the predicted carbon emission output by the LSTM model. The airport terminal carbon emission prediction precision and real-time performance can be remarkably improved, scientific, efficient, real-time and accurate decision support is provided for airport terminal operation management, and low-carbon and near-zero-carbon terminal targets are assisted to be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of building carbon emission prediction, and particularly to a carbon emission prediction method and system for airport terminals. Background Art

[0002] With the continuous increase in the number of civil airports and the steady rise in passenger throughput, the challenges of energy consumption and ecological environment protection faced by airport terminals during operation have become increasingly prominent. Relevant research points out that terminals account for a significant proportion of the overall carbon emissions of airports. Therefore, the energy-saving and carbon-reduction work of airport terminals is undoubtedly an important breakthrough for airports to achieve green and low-carbon development. To promote the high-quality development of airport terminals, accurately predicting the carbon emission level of airport terminals is of great practical significance for airports to formulate low-carbon and near-zero-carbon development plans and clarify the carbon-reduction tasks at each stage.

[0003] The carbon emissions of building systems exhibit random characteristics of non-linearity, dynamics, and uncertainty. In the field of airport terminal carbon emission prediction, traditional prediction methods mostly rely on static energy consumption models. However, these methods often have many deficiencies, such as failing to fully consider the interactions between numerous complex variables in the terminal, the existence of data island phenomena, and the prediction accuracy being difficult to meet actual requirements. In view of this, there is an urgent need to propose a more scientific and accurate carbon emission prediction method and system for airport terminals to effectively improve the accuracy and reliability of airport terminal carbon emission prediction. Summary of the Invention

[0004] The present invention provides a carbon emission prediction method and system for airport terminals, providing technical support for accurate carbon emission prediction and operation management of airport terminals.

[0005] The present invention provides a carbon emission prediction method for airport terminals, including: obtaining key data of the airport terminal at the current moment; inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model calibrated based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on airport terminal carbon emission training samples, and the LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

[0006] According to an airport terminal building carbon emission prediction method provided by the present invention, after inputting the key data of the airport terminal building at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment, it further includes: obtaining the measured carbon emissions at the current moment; calibrating the LSTM-Markov carbon emission prediction model at the current moment according to the predicted carbon emissions at the current moment and the measured carbon emissions at the current moment, and adjusting the parameters of the LSTM-Markov carbon emission prediction model at the current moment in combination with an adaptive learning strategy to obtain the LSTM-Markov carbon emission prediction model for the next moment.

[0007] According to an airport terminal building carbon emission prediction method provided by the present invention, the key data of the airport terminal building includes outdoor environmental data, terminal building operation data, time data, and energy consumption data.

[0008] According to an airport terminal building carbon emission prediction method provided by the present invention, the calculation formula for the carbon emissions of the airport terminal building is: , where, is the total carbon emissions of the terminal building, is the carbon emissions of the i th energy-consuming device in the terminal building, is the consumption of the i th energy-consuming device in the terminal building for the j rd type of energy, is the carbon emission factor for the j th type of energy use, is the electricity generated by photovoltaic power generation in the total electricity consumption of the terminal building, is the power consumption of the APU alternative facility in the terminal building, is the carbon emission factor of electricity, n is the number of energy-consuming devices, k is the number of types of energy.

[0009] A method for predicting carbon emissions of an airport terminal provided by the present invention, the method for obtaining the carbon emission training samples of the airport terminal includes: processing outliers in the key data sample set of the airport terminal by using the Z-score method, and perfecting the key data sample set of the airport terminal by using the method of linear interpolation or average filling to obtain data after outlier processing; performing normalization processing on the data after outlier processing by using the maximum-minimum normalization method to obtain normalized data; according to the normalized data, performing correlation analysis on the key data sample set of the airport terminal and the measured carbon emission sample set by using the Pearson correlation coefficient, and screening out the key data of the airport terminal with a correlation higher than a preset correlation threshold with the airport carbon emissions; performing dimensionality reduction processing on the screened key data of the airport terminal by using the principal component analysis method to obtain the carbon emission training samples of the airport terminal.

[0010] A method for predicting carbon emissions of an airport terminal provided by the present invention further includes: optimizing the LSTM model by using an anti-overfitting algorithm, a hyperparameter optimization algorithm, and a verification algorithm; the anti-overfitting algorithm includes at least one of the Dropout method, the L2 regularization method, and the early stopping method; the hyperparameter optimization algorithm includes the Bayesian optimization algorithm; the verification algorithm includes the nested cross-validation algorithm.

[0011] The Markov model dynamically corrects the predicted carbon emissions output by the LSTM model, including: calculating the relative error of the carbon emissions according to the measured carbon emissions of the airport terminal and the predicted carbon emissions output by the LSTM model; dividing into five Markov state intervals from high to low according to the relative error; updating the state transition probability matrix of the Markov model through Kalman filtering according to the measured carbon emissions of the airport terminal and the predicted carbon emissions output by the LSTM model; calculating the state distribution of the carbon emissions at the future moment according to the Markov state corresponding to the carbon emissions at the current moment and the state transition probability matrix; the state distribution corresponds to one of the five Markov state intervals; correcting the predicted carbon emissions output by the LSTM model according to the state interval corresponding to the state distribution to obtain the predicted carbon emissions output by the Markov model after the state transition probability matrix is updated; the predicted carbon emissions output by the LSTM-Markov carbon emission prediction model is obtained by weighted summing the predicted carbon emissions output by the LSTM model and the predicted carbon emissions output by the Markov model.

[0012] According to an airport terminal carbon emission prediction method provided by the present invention, the LSTM model and the Markov model are trained using the incremental gradient descent algorithm; the parameters of the LSTM model include the number of LSTM layers, the number of hidden units, the Dropout ratio, and the learning rate; the parameters of the Markov model include the state transition probability matrix and the state transition probability matrix update frequency.

[0013] The present invention also provides an airport terminal carbon emission prediction system, including: an acquisition module for acquiring key data of the airport terminal at the current moment; a prediction module for inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emission amount at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model calibrated based on the predicted carbon emission amount at the previous moment and the measured carbon emission amount at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on airport terminal carbon emission training samples, and the LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emission amount output by the LSTM model.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the airport terminal carbon emission prediction method as described in any one of the above.

[0015] An airport terminal carbon emission prediction method and system provided by the present invention acquire key data of the airport terminal at the current moment; input the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emission amount at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model calibrated based on the predicted carbon emission amount at the previous moment and the measured carbon emission amount at the previous moment; the Markov model in the LSTM-Markov carbon emission prediction model dynamically corrects the predicted carbon emission amount output by the LSTM model. The present invention can significantly improve the prediction accuracy and real-time performance of airport terminal carbon emissions, provide scientific, efficient, real-time, and accurate decision-making support for airport terminal operation management, and help achieve the goals of low-carbon and near-zero-carbon terminals. Description of the Drawings

[0016] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0017] Figure 1 It is a schematic flowchart of a method for predicting carbon emissions from an airport terminal building provided by the present invention.

[0018] Figure 2 It is a schematic structural diagram of a system for predicting carbon emissions from an airport terminal building provided by the present invention.

[0019] Figure 3 It is a schematic structural diagram of an electronic device provided by the present invention. Detailed implementation manners

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0021] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for predicting carbon emissions from an airport terminal building provided by the present invention.

[0022] The present invention provides a method for predicting carbon emissions from an airport terminal building, including: 101: Obtain the key data of the airport terminal building at the current moment; 102: Input the key data of the airport terminal building at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; Among them, the LSTM-Markov carbon emission prediction model at the current moment is a model calibrated based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on the carbon emission training samples of the airport terminal building. The LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

[0023] To solve the technical problems existing in the prior art, the present invention provides a method for predicting carbon emissions of an airport terminal. First, key data of the airport terminal at the current moment is obtained. These key data include, but are not limited to, the passenger flow of the airport terminal, the number of flight takeoffs and landings, energy consumption data (such as electricity, fuel, natural gas, etc.), indoor temperature, humidity, equipment operation status, etc. These data can be collected in real time through the airport's intelligent sensor network, energy management system, and flight information system. For example, in a certain airport terminal, the power consumption and equipment operation status are monitored in real time through intelligent electricity meters and sensors installed in various areas; at the same time, the number of flight takeoffs and landings and passenger flow data are obtained from the flight information system.

[0024] The key data of the airport terminal at the current moment collected is input into the LSTM-Markov carbon emission prediction model. The LSTM-Markov carbon emission prediction model is a composite model composed of an LSTM model and a Markov model. The LSTM model is a long short-term memory network that can effectively process time series data and capture the dynamic change law of carbon emissions of the airport terminal. The Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model to further improve the prediction accuracy. Specifically, the LSTM model first extracts features and models the time series of the input key data, and outputs the preliminary predicted carbon emissions. Then, the Markov model corrects the prediction result according to the output result of the LSTM model and the transition probability of the historical carbon emission data. For example, assuming that the LSTM model predicts the carbon emissions at the current moment to be 100 tons, the Markov model judges whether the predicted value may be too high or too low according to the transition probability of the historical data and makes corresponding adjustments, and finally outputs the corrected predicted carbon emissions.

[0025] The LSTM-Markov carbon emission prediction model at the current moment is a model calibrated based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment. Specifically, at each time step, the predicted carbon emissions output by the LSTM-Markov model are compared with the actually measured carbon emissions to calculate the error. According to the error situation, the parameters of the model are adjusted to optimize the prediction performance of the model. For example, if the error between the predicted carbon emissions and the measured carbon emissions is large, the weights of the LSTM model or the transition probability matrix of the Markov model can be adjusted to make the model better adapt to the changes in the actual data.

[0026] The LSTM-Markov carbon emission prediction model is trained based on the training samples of the carbon emissions of the airport terminal building. The training samples include the key data of the airport terminal building at historical moments and the corresponding measured carbon emissions. During the training process, these sample data are used to jointly train the LSTM model and the Markov model, enabling the model to learn the dynamic change rules of the carbon emissions of the airport terminal building and the mutual relationships between different factors. For example, by analyzing historical data, the model can learn the change trend of carbon emissions when the number of flight takeoffs and landings increases, as well as the impact of the change in passenger flow on carbon emissions.

[0027] The method for predicting the carbon emissions of the airport terminal building in this embodiment can be applied to the energy conservation and emission reduction management of the airport. By predicting the carbon emissions in real time, airport managers can formulate energy conservation and emission reduction measures in advance, optimize the energy use strategy, and reduce carbon emissions. For example, according to the prediction results, reasonably adjust the operating parameters of the airport air conditioning system, optimize flight scheduling, and reduce the energy consumption and carbon emissions of the airport. In addition, this method can also provide technical support for the carbon emission monitoring and reporting of the airport, helping the airport meet environmental protection requirements and improve the level of green operation. Through the above embodiments, the method for predicting the carbon emissions of the airport terminal building of the present invention can effectively improve the accuracy and real-time performance of carbon emission prediction, providing strong support for the green and low-carbon development of the airport.

[0028] As a preferred embodiment, after inputting the key data of the airport terminal building at the current moment into the LSTM-Markov carbon emission prediction model at the current moment and obtaining the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment, it further includes: obtaining the measured carbon emissions at the current moment; verifying the LSTM-Markov carbon emission prediction model at the current moment according to the predicted carbon emissions at the current moment and the measured carbon emissions at the current moment, and adjusting the parameters of the LSTM-Markov carbon emission prediction model at the current moment in combination with the adaptive learning strategy to obtain the LSTM-Markov carbon emission prediction model at the next moment.

[0029] In this embodiment, after obtaining the predicted carbon emissions, the measured carbon emissions at the current moment are further obtained. The measured carbon emissions can be obtained through the carbon emission monitoring system of the airport, which is calculated based on real-time energy consumption data and carbon emission factors. For example, the current moment is 9 am, the measured carbon emissions are 118 tons; through the collected key data, the LSTM-Markov model predicts that the carbon emissions at this moment are 120 tons.

[0030] Next, the LSTM-Markov carbon emission prediction model is calibrated based on the predicted carbon emissions and measured carbon emissions at the current moment. Specifically, the error between the predicted carbon emissions and the measured carbon emissions is calculated. It is judged whether the error is within the preset threshold range. If the error exceeds the threshold, it indicates that the model needs to be adjusted. For example, if the error threshold is set to 5 tons and the current error is 2 tons, it means that the model prediction is relatively accurate, but it can still be further optimized. The model parameters are adjusted in combination with the adaptive learning strategy. The adaptive learning strategy can dynamically adjust the model parameters according to the magnitude and direction of the error. For example, if the error is positive, it means that the model prediction is on the high side, and the model prediction value can be reduced by adjusting the weights of the LSTM model or the transition probability matrix of the Markov model. In this example, the error is 2 tons, and the model can fine-tune the parameters to make the prediction value closer to the measured value.

[0031] Through the above calibration and parameter adjustment process, the LSTM-Markov carbon emission prediction model for the next moment is obtained. This model will be used to predict the carbon emissions at the next moment and continue to be calibrated and optimized through the measured data. For example, at 10 am, the updated model is used to predict the carbon emissions to be 125 tons, and the measured carbon emissions are 124 tons, with an error of 1 ton. The model is fine-tuned again according to the error to further improve the prediction accuracy.

[0032] The method in this embodiment is particularly applicable to a complex and dynamically changing environment such as an airport terminal building. By obtaining key data in real time, predicting carbon emissions, calibrating the model, and adaptively adjusting the parameters, the accuracy and real-time performance of carbon emission prediction can be effectively improved. The adaptive learning strategy enables the model to continuously learn and adapt to new data patterns, improving the robustness and adaptability of the model, so as to maintain a high prediction accuracy in long-term operation.

[0033] As a preferred embodiment, the key data of the airport terminal building includes outdoor environmental data, terminal building operation data, time data, and energy consumption data.

[0034] In this embodiment, the outdoor environmental data may include meteorological data, light intensity, and air quality. The meteorological data includes temperature, humidity, wind speed, wind direction, precipitation, etc. These data can be obtained in real time through a meteorological station near the airport. For example, temperature and humidity affect the energy consumption of the air conditioning system in the terminal building, which in turn affects carbon emissions. Light intensity data is crucial for evaluating the energy consumption of the lighting system in the terminal building. Through light sensors, the light intensity inside and outside the terminal building can be monitored in real time, so as to optimize the operation strategy of the lighting system and reduce unnecessary energy consumption. Air quality data (such as PM2.5, PM10, sulfur dioxide, etc.) can reflect the environmental conditions around the airport and have an important impact on the operation of the ventilation system in the terminal building. When the air quality is poor, it may be necessary to increase the operation time of the ventilation system, thereby increasing energy consumption.

[0035] Terminal operation data can include passenger flow, flight takeoff and landing times, and equipment operation status. Real-time passenger flow data is obtained through the airport's security check system, boarding gate statistics system, etc. An increase in passenger flow will lead to an increase in the usage frequency of various facilities in the terminal (such as elevators, escalators, lighting, etc.), thus affecting energy consumption and carbon emissions. Flight takeoff and landing times directly affect the operating energy consumption of the airport, especially the fuel consumption during the taxiing, takeoff, and landing of aircraft. Through the flight information system, the flight takeoff and landing times can be obtained in real time. Equipment operation status includes the operation status of equipment such as the air conditioning system, lighting system, and elevator system in the terminal. Through the equipment management system, the operating parameters of the equipment (such as power, operating time, etc.) can be monitored in real time, so as to evaluate the energy consumption of the equipment.

[0036] Time data can include date and week, timestamp, and season information. There are significant differences in passenger flow and flight takeoff and landing times on different dates and weeks. For example, passenger flow is usually higher on weekends and holidays, and energy consumption also increases accordingly. Specific timestamp data (such as hours, minutes) is crucial for analyzing the daily variation law of carbon emissions. For example, the lighting intensity is higher during the day, and the energy consumption of the lighting system is lower, while more lighting energy consumption is required at night. Seasonal changes have a significant impact on the energy consumption of the terminal. For example, more cooling energy consumption is required in summer, and more heating energy consumption is required in winter.

[0037] Energy consumption data can include electricity consumption, fuel consumption, and natural gas consumption. Through smart meters installed in the terminal, electricity consumption data is monitored in real time. Electricity consumption is one of the main sources of carbon emissions in the terminal, especially the operation of the air conditioning system, lighting system, and various equipment. The fuel consumption of the airport mainly comes from the taxiing, takeoff, and landing processes of aircraft. Through the flight information system and fuel management system, fuel consumption data can be obtained in real time. If natural gas is used as an energy source in the terminal (such as the heating system), natural gas consumption data can be monitored in real time through a natural gas meter.

[0038] In this embodiment, the key data of the airport terminal has significant characteristics in terms of time resolution, spatial resolution, and data diversity.

[0039] In terms of the time resolution characteristics, by deploying high-precision meteorological sensors, the sampling frequency of outdoor environmental data is increased to the millisecond level to ensure the real-time capture of weather changes. For passenger flow data, the fusion technology of millimeter-wave radar and Wi-Fi probes is adopted to generate a second-level dynamic passenger flow density heat map, accurately reflecting the passenger flow situation. At the same time, power quality analyzers are installed on the key energy-consuming equipment in the airport terminal to achieve millisecond-level waveform acquisition of current and voltage, providing high-precision data support for energy management.

[0040] In terms of spatial resolution, weather stations are deployed every 50 meters around the terminal building to form a dense meteorological monitoring network; inside, micro environmental sensors (including temperature, CO 2 concentration, PM2.5, etc.) are deployed in a grid pattern to construct a three-dimensional environmental monitoring network. For special structures such as glass curtain walls and daylighting ceilings, surface temperature and solar radiation sensors are installed, with a spatial density of 1 monitoring point per 10 square meters, achieving precise monitoring and measurement of the terminal building environment. In addition, independent metering modules are installed for the energy-consuming equipment in the terminal building to construct a three-level energy consumption topology structure of equipment - circuit - system; and the terminal building is divided by functional areas, and each sub-area independently collects data such as equipment status, energy consumption, and passenger flow, realizing refined management.

[0041] In terms of data diversity, by leveraging multi-physical field sensing technologies such as sound, light, electricity, and magnetism, combined with multi-modal data fusion methods, the data types are expanded and the data dimensions are enriched. Using the data lake architecture, real-time integration and alignment of multi-source heterogeneous data are achieved, breaking data silos, and providing comprehensive and accurate data support for the intelligent management and decision-making of the airport terminal building.

[0042] In this embodiment, the carbon emission calculation scope of the airport terminal building covers the carbon dioxide emissions generated by energy consumption during the operation of equipment such as heating, ventilation, and air conditioning, lighting systems, domestic hot water supply, catering facilities, elevator systems, baggage handling systems, weak current information systems, and sockets. However, this calculation scope does not include the carbon dioxide emissions corresponding to the energy consumption of APU (auxiliary power unit) replacement facilities, rapid transit systems, and external charging piles.

[0043] As a preferred embodiment, the carbon emission calculation formula for the airport terminal building is: , where, is the total carbon emissions of the terminal building, is the carbon emissions of the i th energy-consuming equipment in the terminal building, is the consumption of the i th energy-consuming equipment in the terminal building for the j th type of energy, is the carbon emission factor for the j th type of energy use, is the electricity generated by photovoltaic power generation in the total electricity consumption of the terminal building, is the electricity consumption of the APU replacement facility in the terminal building, is the carbon emission factor for electricity, n is the number of energy-consuming equipment, k is the number of types of energy.

[0044] As a preferred embodiment, the method for obtaining the carbon emission training samples of the airport terminal building includes: processing the outliers in the key data sample set of the airport terminal building by using the Z-score method, and improving the key data sample set of the airport terminal building by using the method of linear interpolation or mean filling to obtain the data after outlier processing; performing normalization processing on the data after outlier processing by using the maximum-minimum normalization method to obtain the data after normalization processing; according to the data after normalization processing, performing correlation analysis on the key data sample set of the airport terminal building and the measured carbon emission sample set by using the Pearson correlation coefficient, and screening the key data of the airport terminal building with a correlation higher than the preset correlation threshold for the airport carbon emissions; performing dimensionality reduction processing on the screened key data of the airport terminal building by using the principal component analysis method to obtain the carbon emission training samples of the airport terminal building.

[0045] In this embodiment, first, collect the key data sample set of the airport terminal building, which includes outdoor environmental data (such as temperature, humidity, wind speed, etc.), terminal operation data (such as passenger flow, number of flight takeoffs and landings, etc.), time data (such as date, timestamp, etc.) and energy consumption data (such as electricity, fuel, natural gas consumption, etc.). Process the outliers in the data sample set by using the Z-score method. The Z-score method identifies outliers by calculating the Z-score of each data point (i.e., the difference between the data point and the mean divided by the standard deviation). For example, set the threshold of the Z-score to ±3, and the data points outside this range are considered outliers. For the identified outliers, use the method of linear interpolation or mean filling for processing. For example, if the temperature data at a certain moment is abnormal, linear interpolation can be used to fill in the value according to the values of the normal data points before and after this moment; or directly fill in with the mean value of the data points before and after this moment.

[0046] Perform normalization processing on the data after outlier processing by using the maximum-minimum normalization method. This method scales the data to the range of 0 to 1, and the formula is: , where, is the normalized data, X is the original data, is the maximum value in the original data, is the minimum value in the original data.

[0047] Based on the normalized data, the Pearson correlation coefficient is used to analyze the correlation between the key data sample set of the airport terminal and the measured carbon emission sample set. The value range of the Pearson correlation coefficient is [-1, 1]. The closer the value is to 1 or -1, the stronger the correlation. A preset correlation threshold is set, such as 0.5. The key data with a correlation higher than this threshold with the measured carbon emissions is screened out. For example, through analysis, it is found that the correlations between passenger flow, power consumption, and temperature data and carbon emissions are 0.7, 0.8, and 0.6 respectively, all higher than the threshold, so these data are retained for subsequent processing.

[0048] The principal component analysis (PCA, Principal Components Analysis) method is used to perform dimensionality reduction on the screened key data of the airport terminal. PCA reduces the data dimension by projecting the original data into a new feature space and extracting the principal components, while retaining the main features of the data. For example, assume that the screened key data includes 4 features such as passenger flow, power consumption, temperature, and humidity. Through PCA analysis, it is found that the cumulative variance contribution rate of the first two principal components reaches 85%, so the data can be reduced to 2 dimensions. The obtained dimensionality-reduced data is the carbon emission training sample of the airport terminal.

[0049] The method in this embodiment ensures the quality and effectiveness of the training samples through a series of data processing steps. Through outlier processing, the influence of noise in the data on model training is avoided; through normalization processing, data with different dimensions can be uniformly processed; through correlation analysis, data highly correlated with carbon emissions is screened out, improving the prediction accuracy of the model; through dimensionality reduction processing, the data dimension is reduced, improving the training efficiency of the model. This method is particularly suitable for scenarios with complex data and diverse features such as airport terminals. Through high-quality training samples, the performance of the LSTM-Markov carbon emission prediction model can be significantly improved, providing more accurate prediction support for the energy conservation and emission reduction management of the airport, and helping the airport achieve the goal of green and low-carbon development.

[0050] In this embodiment, the LSTM (Long Short-Term Memory) model is a special architecture of the recurrent neural network (RNN, Recurrent Neural Network), which is used to process and predict long-term dependencies in time series data. The core structure of the LSTM model consists of a series of cells, and each cell contains three key gating mechanisms: the forget gate, the input gate, and the output gate. These gating mechanisms work together to control the flow and update of information within the cell, thus effectively solving the problems of vanishing gradients or exploding gradients that traditional RNNs are prone to when dealing with long sequence data.

[0051] The role of the forget gate is to determine which information to discard from the cell state. It does this by looking at the previous hidden state and the current input , and calculating the value of the forget gate through an activation function (usually the sigmoid function). The output of the forget gate is a value between 0 and 1, representing the degree of forgetting for each element in the cell state. If the value of the forget gate is close to 1, it means most of the information is retained; if it is close to 0, it means most of the information is discarded. The calculation formula for the forget gate is: , where, is the forget gate, is the non-linear activation function, is the weight matrix of the forget gate, is the previous hidden state, is the input feature vector, is the bias matrix of the forget gate.

[0052] The role of the input gate is to determine how much of the current input information is written into the cell state. The input gate consists of two parts: one is the activation value of the input gate, and the other is the candidate value. The activation value of the input gate is calculated through the sigmoid function to determine which values will be updated; the candidate value is calculated through the tanh function to generate a new candidate value vector, which will be added to the state. The calculation formula for the input gate is:

[0053] , where, is the input gate, is the weight matrix of the input gate, is the bias matrix of the input gate, is the temporary cell state, is the weight matrix for candidate values, is the bias matrix for candidate values, is to update the cell state, is the cell state at the previous moment.

[0054] The role of the output gate is to determine the value of the next hidden state. It does this by looking at the hidden state at the previous moment and the input at the current moment , and passing them through a sigmoid function to calculate the value of the output gate. The value of the output gate determines how much information in the cell state will be output. The calculation formula for the output gate is: , where, is the output gate, is the weight matrix of the output gate, is the bias matrix of the output gate, is to update the cell state.

[0055] At each time step, the state of the LSTM cell is updated according to the calculation results of the forget gate, input gate, and output gate. The specific update formula is: .

[0056] As a preferred embodiment, it further includes: optimizing the LSTM model using an anti-overfitting algorithm, a hyperparameter optimization algorithm, and a validation algorithm; the anti-overfitting algorithm includes at least one of the Dropout method, the L2 regularization method, and the early stopping method; the hyperparameter optimization algorithm includes the Bayesian optimization algorithm; the validation algorithm includes the nested cross-validation algorithm.

[0057] In this embodiment, a Dropout layer is integrated into the LSTM model to prevent the model from overfitting. The Dropout method randomly discards the outputs of some neurons during training, making the model unable to overly rely on certain specific neurons, thereby enhancing the generalization ability of the model. After using the Dropout method, the performance of the model on the validation set is significantly improved, and the overfitting phenomenon is effectively alleviated.

[0058] L2 regularization is introduced during the training process of the LSTM model. L2 regularization limits the size of the model parameters by adding the L2 norm of the weights to the loss function, thereby simplifying the model and reducing overfitting. L2 regularization can effectively reduce the complexity of the model while improving the performance of the model on the test set.

[0059] Early stopping is adopted during model training. By monitoring the performance on the validation set, training is stopped when the validation set loss no longer decreases or starts to increase, thus avoiding overfitting. Through early stopping, the model can stop in a timely manner during training, avoid overfitting the training data, and at the same time maintain good generalization ability.

[0060] The Bayesian optimization algorithm is used to optimize the hyperparameters of the LSTM model. Bayesian optimization constructs a prior distribution of the objective function and continuously updates the posterior distribution based on the observed data to efficiently search for the optimal combination of hyperparameters. The Bayesian optimization algorithm can efficiently search the hyperparameter space, avoid the high computational cost of traditional grid search or random search, and significantly improve the performance of the model.

[0061] The nested cross-validation algorithm is used to validate the optimized LSTM model. Nested cross-validation includes an inner loop and an outer loop. The inner loop is used for hyperparameter search, and the outer loop is used for model performance evaluation. Through nested cross-validation, overfitting can be effectively avoided, ensuring the reliable performance of the model on an independent test set.

[0062] As a preferred embodiment, the Markov model dynamically corrects the predicted carbon emissions output by the LSTM model, including: calculating the relative error of carbon emissions based on the measured carbon emissions of the airport terminal building and the predicted carbon emissions output by the LSTM model; dividing into five Markov state intervals from high to low according to the relative error; updating the state transition probability matrix of the Markov model through Kalman filtering based on the measured carbon emissions of the airport terminal building and the predicted carbon emissions output by the LSTM model; calculating the state distribution of carbon emissions at the future moment according to the Markov state corresponding to the current moment carbon emissions and the state transition probability matrix; the state distribution corresponds to one of the five Markov state intervals; correcting the predicted carbon emissions output by the LSTM model according to the state interval corresponding to the state distribution to obtain the predicted carbon emissions output by the Markov model after the state transition probability matrix is updated; the predicted carbon emissions output by the LSTM-Markov carbon emission prediction model is obtained by weighted summation of the predicted carbon emissions output by the LSTM model and the predicted carbon emissions output by the Markov model.

[0063] In this embodiment, at each prediction moment, first calculate the relative error between the measured carbon emissions of the airport terminal building and the predicted carbon emissions output by the LSTM model. The calculation formula for the relative error is: , where, is t the relative error between the measured carbon emissions of the airport terminal building and the predicted carbon emissions output by the LSTM model at time is the measured carbon emissions of the airport terminal is the predicted carbon emissions output by the LSTM model

[0064] According to the relative error, the Markov state intervals are divided, and five state intervals such as overestimated, overestimated, relatively accurate, underestimated, and severely underestimated are defined to improve the granularity of Markov state division

[0065] According to the measured carbon emissions of the airport terminal and the predicted carbon emissions output by the LSTM model, it is updated once every 15 minutes, and the state transition probability matrix of the Markov model is updated and optimized through Kalman filtering. The calculation formula is , where is the updated state transition probability matrix is the previous state transition probability matrix is the Kalman gain matrix

[0066] According to the Markov state and state transition probability matrix corresponding to the carbon emissions at the current moment, calculate the state distribution of the carbon emissions at the future moment , where is the state distribution of the carbon emissions at the future moment, and this state distribution corresponds to one of the five Markov state intervals is the Markov state corresponding to the carbon emissions at the current moment

[0067] According to the state interval corresponding to the state distribution, correct the predicted carbon emissions output by the LSTM model to obtain the predicted carbon emissions output by the Markov model after the state transition probability matrix is updated. The calculation formula is

[0068] where is the predicted carbon emissions output by the Markov model

[0069] When the state interval corresponding to the state distribution is in the overestimated or overestimated state interval, the bracket on the right side of the equal sign in the calculation formula of the predicted carbon emissions output by the Markov model takes ; when the state interval corresponding to the state distribution is in the severely underestimated or underestimated state interval, the bracket on the right side of the equal sign in the calculation formula of the predicted carbon emissions output by the Markov model takes ; when the state interval corresponding to the state distribution is in the relatively accurate state interval, the bracket on the right side of the equal sign in the calculation formula of the predicted carbon emissions output by the Markov model takes 1

[0070] The predicted carbon emissions output by the final LSTM-Markov carbon emissions prediction model are obtained by weighted summation of the predicted carbon emissions output by the LSTM model and the predicted carbon emissions output by the Markov model. According to the prediction model error and state stability, the weight factors output by the LSTM model and the Markov model are dynamically optimized, and the calculation formula is: , wherein, is the predicted carbon emissions output by the LSTM-Markov carbon emissions prediction model, is the weight factor of the LSTM model, is the weight factor of the Markov model.

[0071] In this embodiment, the predicted carbon emissions output by the LSTM-Markov carbon emissions prediction model are compared with the measured carbon emissions to evaluate and monitor the performance of the prediction model. In order to verify the prediction accuracy of the prediction model, indicators such as the Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Coefficient of Determination (R 2 ) can be used to quantify the prediction accuracy.

[0072] , wherein, u is the number of sample data, is the predicted carbon emissions, y v is the measured carbon emissions, is the average value of the measured carbon emissions. The smaller the RMSE and MAE, the better the prediction result, and the closer R 2 is to 1, the better the prediction result.

[0073] Deploy edge computing nodes to update the model prediction result error in real time and record it.

[0074] As a preferred embodiment, the LSTM model and the Markov model are trained using the incremental gradient descent algorithm; the parameters of the LSTM model include the number of LSTM layers, the number of hidden units, the Dropout ratio, and the learning rate; the parameters of the Markov model include the state transition probability matrix and the state transition probability matrix update frequency.

[0075] In this embodiment, the Incremental Gradient Descent (IGD) algorithm is adopted to train the LSTM model and the Markov model in real time, and the model parameters are gradually updated using new data. The LSTM model mainly updates key parameters such as the number of LSTM layers, the number of hidden units, the Dropout ratio, and the learning rate; the Markov model mainly updates key parameters such as the state transition matrix and the state transition matrix update frequency.

[0076] The present invention comprehensively considers the temporal and non-linear characteristics of airport terminal carbon emission data, organically integrates the Internet of Things perception network with the intelligent prediction model, significantly improves the model prediction performance, effectively guarantees the accuracy and real-time nature of airport terminal prediction, and provides technical support for the efficient management of airport terminal carbon emissions and scientific carbon reduction.

[0077] The airport terminal carbon emission prediction system provided by the present invention will be described below. The airport terminal carbon emission prediction system described below can be correspondingly referred to the airport terminal carbon emission prediction method described above.

[0078] Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of an airport terminal carbon emission prediction system provided by the present invention.

[0079] The present invention also provides an airport terminal carbon emission prediction system, including: an acquisition module 201, configured to acquire key data of the airport terminal at the current moment; a prediction module 202, configured to input the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment, and obtain the predicted carbon emission amount at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emission amount at the previous moment and the measured carbon emission amount at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on airport terminal carbon emission training samples, and the LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emission amount output by the LSTM model.

[0080] Based on the background of the green and low-carbon development of airport terminals, the present invention proposes an airport terminal carbon emission prediction system, which comprehensively considers the temporal and non-linear characteristics of airport terminal carbon emission data, organically integrates the Internet of Things perception network with the intelligent prediction model, improves the prediction accuracy and robustness of the model, aims to promote the high-quality development of airport terminals, and provides technical support for the accurate prediction and operation management of airport terminal carbon emissions.

[0081] Figure 3 An entity structure schematic diagram of an electronic device is exemplified, such as Figure 3 shown. The electronic device may include: a processor 301, a communications interface 302, a memory 303, and a communication bus 304. Among them, the processor 301, the communications interface 302, and the memory 303 complete mutual communication through the communication bus 304. The processor 301 may call logical instructions in the memory 303 to execute the carbon emission prediction method for an airport terminal building. The method includes: obtaining key data of the airport terminal building at the current moment; inputting the key data of the airport terminal building at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emission amount at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emission amount at the previous moment and the measured carbon emission amount at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on carbon emission training samples of the airport terminal building. The LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emission amount output by the LSTM model.

[0082] In addition, when the logical instructions in the above-mentioned memory 303 are implemented in the form of software functional units and sold or used as an independent product, they may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0083] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the airport terminal carbon emission prediction method provided by the above-mentioned various methods. The method includes: obtaining key data of the airport terminal at the current moment; inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on airport terminal carbon emission training samples. The LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

[0084] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the airport terminal carbon emission prediction method provided by the above-mentioned various methods. The method includes: obtaining key data of the airport terminal at the current moment; inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment to obtain the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; wherein, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on airport terminal carbon emission training samples. The LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

[0085] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0086] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting carbon emissions from an airport terminal, characterized in that: include: Get the key data of the airport terminal at the current moment; Inputting the airport terminal key data at the current moment into the LSTM-Markov carbon emission prediction model at the current moment, and obtaining the predicted carbon emissions at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; Among them, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on the airport terminal carbon emission training samples, and the LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

2. The method for predicting carbon emissions from an airport terminal according to claim 1, characterized in that: After inputting the airport terminal key data at the current moment into the LSTM-Markov carbon emission prediction model at the current moment and obtaining the predicted carbon emission at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment, the method further includes: Get the measured carbon emissions at the current moment; The LSTM-Markov carbon emission prediction model at the current moment is verified according to the predicted carbon emissions at the current moment and the measured carbon emissions at the current moment, and the LSTM-Markov carbon emission prediction model at the current moment is adjusted in combination with an adaptive learning strategy to obtain the LSTM-Markov carbon emission prediction model at the next moment.

3. The method for predicting carbon emissions from an airport terminal according to claim 1, characterized in that: The airport terminal key data includes outdoor environment data, terminal operation data, time data and energy consumption data.

4. The method for predicting carbon emissions from an airport terminal according to claim 1, characterized in that: The formula for calculating carbon emissions from airport terminals is: ， in, is the total carbon emissions of the terminal, The first i Carbon emissions from energy-consuming equipment The first i Energy-consuming equipment j Energy consumption, For the j The carbon emission factor of energy use is The amount of electricity provided by photovoltaic power generation in the total electricity consumption of the terminal building. Power consumption for the terminal APU replacement facilities, is the carbon emission factor of electricity, n is the number of energy-consuming devices, k is the number of energy types.

5. The method for predicting carbon emissions from an airport terminal according to claim 2, characterized in that: The method for obtaining the airport terminal carbon emission training sample includes: The Z-score method is used to process the outliers in the airport terminal key data sample set, and the airport terminal key data sample set is improved by linear interpolation or mean value filling method to obtain the data after outlier processing; Normalizing the abnormally processed data using a maximum-minimum normalization method to obtain normalized data; Based on the normalized data, a Pearson correlation coefficient is used to perform a correlation analysis on a sample set of airport terminal key data and a sample set of measured carbon emissions, and the key data of airport terminals whose correlation with airport carbon emissions is higher than a preset correlation threshold are screened; The principal component analysis method is used to reduce the dimension of the screened key data of the airport terminal to obtain the carbon emission training samples of the airport terminal.

6. The method for predicting carbon emissions from an airport terminal according to claim 1, characterized in that: Also includes: The LSTM model is optimized by using an anti-overfitting algorithm, a hyperparameter optimization algorithm and a verification algorithm; the anti-overfitting algorithm includes at least one of a Dropout method, an L2 regularization method and an early stopping method; the hyperparameter optimization algorithm includes a Bayesian optimization algorithm; and the verification algorithm includes a nested cross-validation algorithm.

7. The method for predicting carbon emissions from an airport terminal according to claim 1, characterized in that: The Markov model dynamically corrects the predicted carbon emissions output by the LSTM model, including: Calculate the relative error of carbon emissions based on the measured carbon emissions of the airport terminal and the predicted carbon emissions output by the LSTM model; According to the relative error, it is divided into five Markov state intervals from high to low; According to the measured carbon emissions of the airport terminal and the predicted carbon emissions output by the LSTM model, the state transition probability matrix of the Markov model is updated through Kalman filtering; Calculate the state distribution of carbon emissions at a future time according to the Markov state corresponding to the carbon emissions at a current time and the state transition probability matrix; the state distribution corresponds to one state interval of the five Markov state intervals; According to the state interval corresponding to the state distribution, the predicted carbon emissions output by the LSTM model are corrected to obtain the predicted carbon emissions output by the Markov model after the state transition probability matrix is ​​updated; The predicted carbon emissions output by the LSTM-Markov carbon emissions prediction model is obtained by weighted summing the predicted carbon emissions output by the LSTM model and the predicted carbon emissions output by the Markov model.

8. The method for predicting carbon emissions from an airport terminal according to any one of claims 1 to 7, characterized in that: The LSTM model and the Markov model are trained using an incremental gradient descent algorithm; the parameters of the LSTM model include the number of LSTM layers, the number of hidden units, the Dropout ratio and the learning rate; the parameters of the Markov model include the state transition probability matrix and the state transition probability matrix update frequency.

9. An airport terminal carbon emission prediction system, characterized in that: include: The acquisition module is used to obtain the key data of the airport terminal at the current moment; A prediction module, used for inputting the key data of the airport terminal at the current moment into the LSTM-Markov carbon emission prediction model at the current moment, and obtaining the predicted carbon emission at the current moment output by the LSTM-Markov carbon emission prediction model at the current moment; Among them, the LSTM-Markov carbon emission prediction model at the current moment is a model verified based on the predicted carbon emissions at the previous moment and the measured carbon emissions at the previous moment; the LSTM-Markov carbon emission prediction model is trained based on the airport terminal carbon emission training samples, and the LSTM-Markov carbon emission prediction model includes an LSTM model and a Markov model, and the Markov model is used to dynamically correct the predicted carbon emissions output by the LSTM model.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method for predicting carbon emissions of an airport terminal as claimed in any one of claims 1 to 8 is implemented.

Images