Carbon footprint prediction method and system based on long and short term cycle network

Abstract

The invention discloses a carbon footprint prediction method and system based on a long and short term cycle network. The method comprises the following steps: acquiring full-sample carbon footprint data; using a traditional LSTM model to predict the full-sample carbon footprint data, and obtaining a sample at an abnormal time point of the LSTM model when prediction is abnormal; labeling operation is carried out on the samples at the abnormal time points, feature extraction is carried out on the labeled samples at the abnormal time points by using a logistic regression method, features influencing the abnormity of the LSTM model are extracted, and a specific feature set is constructed; constructing a first-stage carbon prediction model based on the specific feature set; using an LSTM (Long Short Term Memory) model to re-learn the residual error of the first-stage carbon prediction model, and determining the re-learned model as a second-stage carbon prediction model; superposing the first-stage carbon prediction model and the second-stage carbon prediction model to obtain a carbon footprint prediction model; and performing carbon footprint prediction by using the carbon footprint prediction model. According to the scheme, the accuracy of carbon footprint prediction can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission prediction, and in particular to a carbon footprint prediction method and system based on a long- and short-term cycle network. Background Art

[0002] Carbon footprint refers to carbon emissions. The main sources of urban carbon emissions include transportation, industrial production, building energy consumption, and residents' lives, which have significant temporal and spatial heterogeneity and multi-scale variation characteristics. However, due to the dynamic characteristics and complexity of urban carbon emissions, existing technologies still have many deficiencies in the accurate prediction and regulation of carbon emission hotspots.

[0003] AI-based carbon footprint forecasting refers to the use of artificial intelligence technology to analyze and predict future carbon footprint trends. These systems typically use machine learning algorithms to identify patterns and factors related to carbon footprints from large amounts of historical data. Through these patterns, AI can predict carbon footprints within a specific time period in the future.

[0004] The traditional LSTM carbon footprint prediction method is limited by the model prediction itself and can only identify information in past time data, thereby mining a periodic pattern. When emergencies occur (such as holidays, major documents, important events, etc.), the model's prediction ability often drops significantly. The fundamental reason is that the model itself does not take into account the impact of this information, thus affecting the accuracy of carbon footprint prediction. Summary of the invention

[0005] In order to solve the technical problem of low prediction accuracy of the traditional LSTM carbon footprint prediction method, the embodiment of the present invention provides a carbon footprint prediction method and system based on long-term and short-term recurrent networks.

[0006] The technical solution of the embodiment of the present invention is achieved as follows: The embodiment of the present invention provides a carbon footprint prediction method based on a long-short term recurrent network, and the method includes: obtaining carbon footprint data of all samples; using a traditional LSTM model to predict the carbon footprint data of all samples, and obtaining samples at abnormal time points of the LSTM model when an abnormality is predicted; labeling the samples at the abnormal time points, and using a logistic regression method to extract features of the labeled samples at the abnormal time points, extracting features that affect the abnormality of the LSTM model, and constructing a specific feature set; constructing a one-stage carbon prediction model based on the specific feature set; and using the LSTM model to re-learn the residual of the one-stage carbon prediction model, and determining the re-learned model as a two-stage carbon prediction model; superimposing the one-stage carbon prediction model and the two-stage carbon prediction model to obtain a carbon footprint prediction model; and using the carbon footprint prediction model to perform carbon footprint prediction.

[0007] In one embodiment, a traditional LSTM model is used to predict the carbon footprint data of the entire sample, and samples at the abnormal time point of the LSTM model when the prediction is abnormal are obtained, including: using a traditional LSTM model to predict the carbon footprint data of the entire sample, and obtaining the prediction residual and relative error predicted by the LSTM model; based on the relative error and a preset threshold, determining whether the prediction of the LSTM model is abnormal; and obtaining samples at the abnormal time point of the LSTM model when the prediction is abnormal.

[0008] In one embodiment, obtaining the prediction residual and relative error predicted by the LSTM model includes: obtaining the prediction residual and relative error predicted by the LSTM model using the following calculation formula: in, is the prediction residual, is the relative error, represents the true value, represents the fitted value, and i represents an individual.

[0009] In one embodiment, based on the relative error and the preset threshold, determining whether the LSTM model prediction is abnormal includes: based on the relative error and the preset threshold, determining whether the LSTM model prediction is abnormal using the following calculation formula: in, is the discriminant function, , model abnormality; , the model is normal; is the relative error, is the preset threshold.

[0010] In one embodiment, a logistic regression method is used to extract features of the samples at the abnormal time point after labeling, and features that affect the abnormality of the LSTM model are extracted, including: Using a logistic regression model to process the samples at the abnormal time point after labeling to obtain a significance level; The features whose significance level is not greater than the preset threshold are eliminated to obtain the features that affect the abnormality of the LSTM model.

[0011] In one embodiment, constructing a one-stage carbon prediction model based on the specific feature set includes: the expression of the one-stage carbon prediction model constructed based on the specific feature set is: in, represents a one-stage carbon prediction model, represents the estimated parameters of the logistic regression model, represents a specific feature set, express The model residuals obtained from the model.

[0012] In one embodiment, the first-stage carbon prediction model and the second-stage carbon prediction model are superimposed to obtain a carbon footprint prediction model, including: using the following calculation formula to superimpose the first-stage carbon prediction model and the second-stage carbon prediction model to obtain a carbon footprint prediction model: Among them, C is the carbon footprint prediction model, represents a one-stage carbon prediction model, Represents a two-stage carbon prediction model.

[0013] An embodiment of the present invention also provides a carbon footprint prediction system based on a long-term and short-term cycle network, comprising: a processor and a memory for storing a computer program that can be run on the processor; wherein the processor executes the steps of the above-mentioned method when running the computer program.

[0014] An embodiment of the present invention further provides a storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0015] An embodiment of the present invention also provides a computer program product, which includes computer instructions. The computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps of the above-mentioned method.

[0016] This embodiment has the following beneficial effects: This embodiment is improved on the basis of the LSTM model. By analyzing and mining the abnormal time points predicted by the model, the logistic regression method is combined to identify the features that may cause influence in advance, and the OLS regression method is used to control the influence of these features in advance, and then the residual is re-learned using the LSTM neural network, and the carbon footprint prediction is obtained by superimposing the two parts of the prediction. The scheme of this embodiment can significantly improve the accuracy of carbon prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a carbon footprint prediction method based on a long-term and short-term cycle network according to an embodiment of the present invention; Figure 2This is a schematic diagram of the LSTM long short-term neural network architecture according to an embodiment of the present invention; Figure 3 1 is a diagram showing the internal structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

[0019] The embodiment of the present invention provides a carbon footprint prediction method based on long-term and short-term cycle networks, such as Figure 1 As shown, the method includes: Step 101: Obtain carbon footprint data of the entire sample; Step 102: Use the traditional LSTM model to predict the carbon footprint data of the entire sample, and obtain samples at the abnormal time point of the LSTM model when the prediction is abnormal; Step 103: labeling the samples at the abnormal time point, and extracting features of the labeled samples at the abnormal time point using a logistic regression method, extracting features that affect the abnormality of the LSTM model, and constructing a specific feature set; Step 104: constructing a one-stage carbon prediction model based on the specific feature set; and relearning the residual of the one-stage carbon prediction model using the LSTM model, and determining the relearned model as the two-stage carbon prediction model; Step 105: superimpose the first-stage carbon prediction model and the second-stage carbon prediction model to obtain a carbon footprint prediction model; and use the carbon footprint prediction model to perform carbon footprint prediction.

[0020] Before introducing the solution of this embodiment, a brief introduction to the traditional LSTM model is given.

[0021] See also Figure 2 , the input of the LSTM gate unit is the carbon footprint data input of the current time step The hidden state at the previous time step , the output is calculated by a fully connected layer whose activation function is the sigmoid function.

[0022] Specifically, assuming the number of hidden units is , given a mini-batch of carbon footprint data input at time step t (The number of samples is , the number of inputs is ) and the hidden state of the previous time step . Time step Input gate , Forget Gate , output gate , candidate memory cells and memory cells The calculations are as follows: (1) (2) (3) (4) (5) (6) in, , , , and , , , is the weight parameter, , , , is the bias parameter, symbol is the element-wise multiplication symbol, and Represent the sigmoid function and the hyperbolic tangent function respectively, and their specific formulas are as follows: (7) (8) Input Gate Used to control the input, transform the input features through matrix operations and sigmoid function operations for further calculations. The sigmoid function is applied to the input to obtain a value of [0,1] and multiply it with the information stored in the memory cell of the previous time step to obtain the retained information, thereby forgetting the previous state information. Output gate Used to control the output, combining the current input features and the hidden state of the previous time step through the sigmoid function to get the output result of the current time step. It is used to record the input feature information of the current time step and is connected to the input gate to determine the new input information into the memory cell. The memory cell of the previous time step is The information in is forgotten, and the input gate retains the information of the current time step to achieve update. Hidden state The updating is achieved by controlling the information of the memory cell through the output gate.

[0023] Based on the traditional LSTM model, this embodiment has been improved on the basis of the traditional LSTM model. By analyzing and mining the abnormal time points predicted by the model, the features that may cause influence are identified in advance by combining the logistic regression method, and the OLS regression method is used to control the influence of these features in advance, and then the residual is re-learned using the LSTM neural network, and the carbon footprint prediction is obtained by superimposing the two parts of the prediction. The scheme of this embodiment can significantly improve the accuracy of carbon prediction.

[0024] Specifically, the content of this embodiment includes: 1. Use the carbon emission data of the entire sample to build an LSTM model for pre-training. The model is recorded as pre-LSTM. The prediction residual of the model is obtained , and the relative error The prediction residual of the model is obtained by subtracting the true value from the pre-LSTM prediction value, and the relative error is obtained by dividing the residual by the true value. The specific formula is as follows: in, represents the true value, represents the fitted value, and i represents an individual.

[0025] 2. Determine whether the model's prediction on the training sample is abnormal, and extract the sample at the time when the model is abnormal. Set the threshold , the discriminant function is ,when hour, , the model is abnormal; otherwise, , the model is normal. The formula is as follows: Extract the time points when the model is abnormal and construct a new sample set S.

[0026] 3. Classify and label the samples in the sample set S, for example, whether a sample is a specific holiday or has any emergency events, and construct a label set SL. The features in the label set are 01 virtual variables. For example, if a sample is a specific holiday, then the label feature is 1, otherwise it is 0.

[0027] 4. Determine which features in the label set will affect the model's anomalies, and extract these features. This example uses sample S to perform Logistic regression, and the model is as follows: in, is an exponential function, is the parameter obtained by Logistic. The significance level of the features in SL is obtained through Logistic regression. When the significance level exceeds 5%, it is considered that the feature can significantly affect the probability of model abnormality. The features that do not reach the 5% significance level are eliminated, and finally the feature set ESL that has a significant impact on model abnormality is obtained.

[0028] 5. Use all samples for training and build a regression model: in, represents a one-stage carbon prediction model, denotes the estimated parameters of logistic regression, represents the specific label vector set, express The model residuals obtained from the model.

[0029] 6. Use LSTM model to The residual error obtained from the model , re-learning, the input is the carbon footprint data on the time series, and the LSTM re-learning model for the residual is recorded as .

[0030] 7. Get the final carbon footprint prediction model.

[0031] This embodiment targets the specific characteristics of personal carbon emissions, that is, fluctuations different from normal times will occur at specific time points, such as during specific holidays or emergencies. Therefore, improvements are made on the basis of the LSTM model. By identifying the time points in advance, after controlling the specificity of consumption carbon emissions in the time series, the residual is re-learned using the LSTM neural network, and the carbon footprint prediction is obtained by superimposing the two parts of the prediction.

[0032] In order to implement the method of an embodiment of the present invention, an embodiment of the present invention also provides a carbon footprint prediction system based on a long-term and short-term cycle network, including: a processor and a memory for storing a computer program that can be run on the processor; wherein, when the processor is used to run the computer program, the steps of the above-mentioned method are executed.

[0033] The above-mentioned system provided in this embodiment belongs to the same concept as the above-mentioned method embodiment. The specific implementation process thereof is detailed in the method embodiment and will not be repeated here.

[0034] In order to implement the method of the embodiment of the present invention, the embodiment of the present invention also provides a computer program product, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of the above method.

[0035] Based on the hardware implementation of the above program modules, and in order to implement the method of the embodiment of the present invention, the embodiment of the present invention further provides an electronic device (computer device). Specifically, in one embodiment, the computer device may be a terminal, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected through a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, the method of any one of the above embodiments is implemented. The display screen A04 of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0036] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0037] The device provided by the embodiment of the present invention includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the method of the above embodiment is implemented.

[0038] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0039] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0040] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0041] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0042] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0043] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0044] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0045] It can be understood that the memory of the embodiment of the present invention can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM, SyncLink Dynamic Random Access Memory), and direct RAMbus random access memory (DRRAM, Direct Rambus Random Access Memory).The memories described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memories.

[0046] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0047] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A carbon footprint prediction method based on long-term and short-term cycle networks, characterized in that: The method comprises: Obtain carbon footprint data for the entire sample; Use the traditional LSTM model to predict the carbon footprint data of the entire sample, and obtain samples at the abnormal time point of the LSTM model when the prediction is abnormal; Labeling the samples at the abnormal time point, and using a logistic regression method to extract features of the labeled samples at the abnormal time point, extracting features that affect the abnormality of the LSTM model, and constructing a specific feature set; Constructing a one-stage carbon prediction model based on the specific feature set; and relearning the residual of the one-stage carbon prediction model using the LSTM model, and determining the relearned model as a two-stage carbon prediction model; The first-stage carbon prediction model and the second-stage carbon prediction model are superimposed to obtain a carbon footprint prediction model; and the carbon footprint prediction model is used to perform carbon footprint prediction.

2. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 1 is characterized in that: Use the traditional LSTM model to predict the carbon footprint data of the entire sample, and obtain samples at the abnormal time point of the LSTM model when the prediction is abnormal, including: Use a traditional LSTM model to predict the carbon footprint data of the entire sample, and obtain the prediction residual and relative error predicted by the LSTM model; Based on the relative error and a preset threshold, determining whether the LSTM model prediction is abnormal; Obtain samples at the abnormal time point of the LSTM model when predicting an abnormality.

3. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 2 is characterized in that: Obtain the prediction residual and relative error predicted by the LSTM model, including: Use the following calculation formula to obtain the prediction residual and relative error predicted by the LSTM model: in, is the prediction residual, is the relative error, represents the true value, represents the fitted value, and i represents an individual.

4. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 2 is characterized in that: Based on the relative error and a preset threshold, determining whether the LSTM model prediction is abnormal includes: Based on the relative error and the preset threshold, the following calculation formula is used to determine whether the LSTM model prediction is abnormal: in, is the discriminant function, , model abnormality; , the model is normal; is the relative error, is the preset threshold.

5. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 1 is characterized in that: The features of the samples at the abnormal time point after labeling are extracted using a logistic regression method to extract the features that affect the abnormality of the LSTM model, including: Using a logistic regression model to process the samples at the abnormal time point after labeling to obtain a significance level; The features whose significance level is not greater than the preset threshold are eliminated to obtain the features that affect the abnormality of the LSTM model.

6. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 1 is characterized in that: A first-stage carbon prediction model is constructed based on the specific feature set, including: The expression of the one-stage carbon prediction model constructed based on the specific feature set is: in, represents a one-stage carbon prediction model, represents the estimated parameters of the logistic regression model, represents a specific feature set, express The model residuals obtained from the model.

7. The carbon footprint prediction method based on long-term and short-term cycle networks according to claim 1 is characterized in that: The first-stage carbon prediction model and the second-stage carbon prediction model are superimposed to obtain a carbon footprint prediction model, including: The first-stage carbon prediction model and the second-stage carbon prediction model are superimposed using the following calculation formula to obtain a carbon footprint prediction model: Among them, C is the carbon footprint prediction model, represents a one-stage carbon prediction model, Represents a two-stage carbon prediction model.

8. A carbon footprint prediction system based on long-term and short-term cycle networks, characterized in that: include: A processor and a memory for storing a computer program that can be run on the processor; wherein, when the processor is used to run the computer program, the steps of the method according to any one of claims 1 to 7 are executed.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product includes computer instructions, which are stored in a computer-readable storage medium. The processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of the method described in any one of claims 1 to 7.

Images