Carbon emission determination method and device, electronic equipment and nonvolatile storage medium

By generating system behavior sequences and first correlation sequences, and filtering out second correlation sequences with a high degree of influence, and combining a grey model with univariate probability time series and time lag characteristics, the problem of low accuracy in carbon emission prediction is solved, and higher computational accuracy and model generalization ability are achieved.

CN119005747APending Publication Date: 2024-11-22CHINA TELECOM CORP LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411068204.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-22

Smart Images

  • Figure CN119005747A_ABST
    Figure CN119005747A_ABST
Patent Text Reader

Abstract

The invention discloses a carbon emission determination method and device, electronic equipment and a nonvolatile storage medium. The method comprises the following steps: generating a system behavior sequence and a first correlation sequence according to historical carbon emission data and historical data of carbon emission related factors; determining a correlation parameter between the system behavior sequence and each first correlation sequence, and determining the first correlation sequence of which the correlation parameter is greater than a preset correlation threshold as a second correlation sequence; according to the second correlation sequence, a third correlation sequence corresponding to a target time period is determined, and the target time period is a time period located after a historical time period corresponding to the second correlation sequence; and determining carbon emission data corresponding to the target time period according to the third correlation sequence. The technical problem that the accuracy of carbon emission prediction is low due to the fact that a data set of influence factors which can be referred to by carbon emission is limited and an exponential decomposition method is mostly adopted when carbon emission is predicted in related technologies is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of carbon emission prediction, and in particular, to a method, device, electronic device and non-volatile storage medium for determining carbon emission. Background Art

[0002] In order to cope with global climate change, the world is actively promoting the reduction of greenhouse gas emissions, especially carbon dioxide. Promoting the development of low-carbon energy and implementing energy-saving and emission reduction measures have become urgent challenges to be solved. For the calculation and prediction of carbon emissions, there are limited data sets of influencing factors that can be referenced for carbon emissions, and related technologies mostly use exponential decomposition methods when predicting carbon emissions, resulting in certain errors in carbon emissions prediction and low accuracy.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a method, device, electronic device and non-volatile storage medium for determining carbon emissions, so as to at least solve the technical problem that the accuracy of carbon emissions prediction is low due to the limited data set of influencing factors that can be referenced for carbon emissions and the fact that related technologies mostly use exponential decomposition method when predicting carbon emissions.

[0005] According to one aspect of an embodiment of the present application, a method for determining carbon emissions is provided, including: generating a system behavior sequence and a first correlation sequence based on historical carbon emission data and historical data of carbon emission-related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first correlation sequence is a time series of historical values ​​of factors related to carbon emissions; determining correlation parameters between the system behavior sequence and each first correlation sequence, and determining a first correlation sequence whose correlation parameter is greater than a preset correlation threshold as a second correlation sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; determining a third correlation sequence corresponding to a target time period based on the second correlation sequence, wherein the third correlation sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second correlation sequence; determining carbon emission data corresponding to the target time period based on the third correlation sequence.

[0006] Optionally, the length of the first correlation sequence is greater than the length of the system behavior sequence; determining the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as the second correlation sequence includes: generating a time-lag sequence corresponding to the first correlation sequence, wherein the length of each time-lag sequence is the same as the length of the system behavior sequence, and each time-lag sequence corresponds to a lag period, and the lag period is used to characterize the lag between the first correlation sequence corresponding to the time-lag sequence and the system behavior sequence in the time dimension; determining the correlation parameter based on the time-lag sequence and the system behavior sequence; when the lag period corresponding to the time-lag sequence characterizes that the first correlation sequence lags behind the system behavior sequence, and the correlation parameter is greater than the preset correlation threshold, determining the first correlation sequence corresponding to the time-lag sequence as the second correlation sequence.

[0007] Optionally, determining the third related sequence corresponding to the target time period based on the second related sequence includes: determining a feature sequence corresponding to the second related sequence, and standardizing the feature sequence; encoding the position information of each element in the standardized feature sequence through sine and cosine position coding to obtain a position code; based on the position code, mapping the feature sequence to a decoder of a univariate probability time series model, and using the decoder to perform autoregressive prediction based on the feature sequence to obtain a third related sequence, wherein a t distribution head is used in the univariate probability time series model to determine the probability distribution of the result output by the decoder.

[0008] Optionally, standardizing the feature sequence includes: determining the mean and variance of the elements in the feature sequence, and standardizing each element in the feature sequence based on the mean and variance; determining the root mean square corresponding to the elements in the feature sequence after standardization, and normalizing each element in the feature sequence after standardization based on the root mean square, wherein the normalization is used to map each element to the same scale range.

[0009] Optionally, the method also includes: determining a first-order cumulative generation sequence corresponding to the second related sequence, wherein the kth element in the first-order cumulative generation sequence corresponding to the second related sequence is the cumulative sum of the 1st to kth elements in the second related sequence, and k is a positive integer; determining a neighboring mean generation sequence corresponding to the system behavior sequence, wherein the kth element in the neighboring mean generation sequence is the mean of the kth element and the k-1th element in the first-order cumulative generation sequence corresponding to the system behavior sequence; determining a time-lag gray model based on the first-order cumulative generation sequence corresponding to the second related sequence, the neighboring mean generation sequence corresponding to the system behavior sequence, and the lag period corresponding to the second related sequence, wherein the time-lag gray model is used to characterize the correlation between the carbon emission-related factors represented by the second related sequence and the carbon emissions.

[0010] Optionally, determining the carbon emission data corresponding to the target time period according to the third related sequence includes: using a time-lag grey model to determine the carbon emission data corresponding to the target time period according to the third related sequence.

[0011] Optionally, generating a system behavior sequence and a first related sequence based on historical carbon emission data and historical data of carbon emission related factors includes: acquiring historical carbon emission data and historical data of carbon emission related factors, wherein the historical carbon emission data includes at least one of the following: carbon emission data of previous years, and the historical data of carbon emission related factors includes at least one of the following: total population of previous years, per capita GDP, total consumption of coal, natural gas and oil, and total value of secondary industry; generating a system behavior sequence based on carbon emission data of previous years; generating a first related sequence based on historical data of carbon emission related factors, wherein each carbon emission related factor corresponds to a first related sequence.

[0012] According to another aspect of an embodiment of the present application, a carbon emission determination device is also provided, including: a data sequence acquisition module, used to generate a system behavior sequence and a first related sequence based on historical carbon emission data and historical data of carbon emission-related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first related sequence is a time series of historical values ​​of factors related to carbon emissions; a strong correlation factor determination module, used to determine the correlation parameters between the system behavior sequence and each first related sequence, and determine the first related sequence whose correlation parameter is greater than a preset correlation threshold as a second related sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; a univariate prediction module, used to determine a third related sequence corresponding to a target time period based on the second related sequence, wherein the third related sequence is a sequence of numerical values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second related sequence; a carbon emission determination module, used to determine the carbon emission data corresponding to the target time period based on the third related sequence.

[0013] According to another aspect of an embodiment of the present application, there is also provided an electronic device, including: a memory and a processor, the processor being configured to run a program stored in the memory, wherein the carbon emission determination method is executed when the program is run.

[0014] According to another aspect of the embodiments of the present application, a non-volatile storage medium is provided, the non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the carbon emission determination method by running the computer program.

[0015] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, which implements the steps of the carbon emission determination method when the computer program is executed by a processor.

[0016] In an embodiment of the present application, a system behavior sequence and a first correlation sequence are generated based on historical carbon emission data and historical data of carbon emission-related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first correlation sequence is a time series of historical values ​​of factors related to carbon emissions; a correlation parameter between the system behavior sequence and each first correlation sequence is determined, and a first correlation sequence whose correlation parameter is greater than a preset correlation threshold is determined as a second correlation sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; based on the second correlation sequence, a third correlation sequence corresponding to the target time period is determined, wherein the third correlation sequence is a time series of historical values ​​of factors related to carbon emissions. The values ​​of the related factors are arranged in chronological order within the target time period, and the target time period is the time period after the historical time period corresponding to the second related sequence; according to the third related sequence, the method of determining the carbon emission data corresponding to the target time period is to start from the univariate time series and predict the factors strongly correlated with carbon emissions, and then use the model with time lag characteristics combined with the univariate predicted data to calculate the carbon emissions, thereby achieving the purpose of improving the accuracy of carbon emission calculation, and further solving the technical problem of low accuracy of carbon emission prediction due to the limited data set of influencing factors that can be used for reference of carbon emissions and the fact that most related technologies use exponential decomposition method when predicting carbon emissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 It is a hardware structure block diagram of a computer terminal (or electronic device) for implementing a method for determining carbon emissions according to an embodiment of the present application;

[0019] Figure 2 It is a schematic diagram of a method flow for determining carbon emissions provided in an embodiment of the present application;

[0020] Figure 3 It is a schematic diagram of a process architecture of a carbon emission prediction method provided according to an embodiment of the present application;

[0021] Figure 4 A schematic diagram of a method flow for predicting carbon emissions based on a univariate probabilistic time series prediction model and a time lag feature GM(1,N) provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of the structure of a univariate probability time series model provided according to an embodiment of the present application;

[0023] Figure 6 A schematic diagram of a carbon emission prediction effect provided according to an embodiment of the present application;

[0024] Figure 7 It is a structural schematic diagram of a carbon emission determination device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] In order to facilitate those skilled in the art to better understand the embodiments of the present application, some technical terms or nouns involved in the embodiments of the present application are explained as follows:

[0028] Lag-Llama (Towards Foundation Models for Probabilistic Time Series Forecasting): is a transformer-based pure decoder model, a general variable probabilistic time series forecasting model trained on a large amount of time series data.

[0029] GM(1, N): A grey model with N variables in a first-order equation. It is a multivariable prediction model in grey system theory and is suitable for processing systems with multiple variables.

[0030] RMS Norm (Root Mean Square Layer Normalization): It is a type of normalization operation that normalizes the tensor to Gaussian distribution and improves the training efficiency and stability of the model by standardizing the input vector.

[0031] Sinusoidal (Sinusoidal Positional Encoding): is an encoding method for sequence data that provides a unique representation for each position in the sequence.

[0032] There are three main emission prediction methods in related technologies: one is based on time series prediction technology, using historical data for prediction and fitting polynomial expressions, but there are certain requirements for the amount of data. The second is to predict carbon emissions based on a neural network model, normalize historical carbon emissions before preprocessing, and then predict through a neural network. It is not possible to simulate and predict small-scale data well. The third is based on a multivariate prediction system. Due to the complexity of the carbon emission system, it is affected by factors such as population, economy, and energy. The prediction of this method mainly uses the exponential decomposition method, which has uncertainty in the prediction results.

[0033] In order to solve the above problems, relevant solutions are provided in the embodiments of the present application, which are described in detail below.

[0034] According to an embodiment of the present application, a method embodiment for determining carbon emissions is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0035] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal (or electronic device) for implementing a method for determining carbon emissions is shown. Figure 1 As shown, the computer terminal 10 (or electronic device) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0036] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or electronic device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0037] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the carbon emission determination method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the above-mentioned carbon emission determination method is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0038] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0039] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or electronic device).

[0040] Under the above operating environment, the embodiment of the present application provides a method for determining carbon emissions. Figure 2 is a schematic diagram of a method flow for determining carbon emissions provided in an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0041] Step S202, generating a system behavior sequence and a first correlation sequence based on historical carbon emission data and historical data of carbon emission related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first correlation sequence is a time series of historical values ​​of factors related to carbon emissions;

[0042] Step S204, determining the correlation parameter between the system behavior sequence and each first correlation sequence, and determining the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as the second correlation sequence, wherein the correlation parameter is used to characterize the influence of the carbon emission related factors on the carbon emission;

[0043] Step S206, determining a third correlation sequence corresponding to the target time period based on the second correlation sequence, wherein the third correlation sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second correlation sequence;

[0044] Step S208: determining the carbon emission data corresponding to the target time period according to the third correlation sequence.

[0045] Through the above steps, starting from the univariate time series, predictions are made for factors strongly correlated with carbon emissions, and then the carbon emissions are calculated using a model with time lag characteristics combined with the univariate predicted data, thereby achieving the purpose of improving the accuracy of carbon emissions calculations. This solves the technical problem of low accuracy in carbon emissions prediction due to the limited data set of influencing factors that can be used as a reference for carbon emissions, and the fact that most related technologies use exponential decomposition methods when predicting carbon emissions.

[0046] The following further introduces the method for determining the carbon emissions in steps S202 to S208 of the embodiment of the present application.

[0047] like Figure 3As shown, first collect and process the original data, that is, the historical carbon emission data and carbon emission related factors. In this embodiment, the carbon emission related factors may include: total population X2, per capita GDP X3, total energy consumption (coal consumption, oil consumption, natural gas consumption) X4, and the total value of the secondary industry X5; then, the GM (1, N) model with a time lag feature can be used to calculate the strong correlation between the related factors and carbon emissions, that is, the first related sequence with a correlation parameter greater than a preset correlation threshold is determined as the second related sequence, and then the carbon emission related factors (i.e., the second related factors) can be predicted based on the Lag-Llama time series model, that is, the third related sequence corresponding to the target time period is determined, and finally the carbon emissions are predicted by the third related sequence predicted by the GM (1, N) model with a time lag feature. The above process architecture is further introduced in detail below.

[0048] Figure 4 A schematic diagram of a method flow for predicting carbon emissions based on a univariate probability time series prediction model and a time lag feature GM(1,N) provided in an embodiment of the present application is shown as follows: Figure 4 shown.

[0049] First, in the data preparation and preprocessing stage, historical carbon emission data and historical data of carbon emission related factors are obtained to generate system behavior sequences and first related sequences. The specific steps are as follows.

[0050] In some embodiments of the present application, generating a system behavior sequence and a first related sequence based on historical carbon emission data and historical data of carbon emission related factors includes the following steps: acquiring historical carbon emission data and historical data of carbon emission related factors, wherein the historical carbon emission data includes at least one of the following: carbon emission data over the years, and the historical data of carbon emission related factors includes at least one of the following: total population over the years, per capita GDP, total coal, natural gas and oil consumption, and total value of the secondary industry; generating a system behavior sequence based on the carbon emission data over the years; generating a first related sequence based on the historical data of carbon emission related factors, wherein each carbon emission related factor corresponds to a first related sequence.

[0051] Specifically, obtain the historical carbon emission data, total population, per capita GDP, total coal, natural gas and oil consumption, and total value of the secondary industry of a certain region, and set the system behavior sequence based on the carbon emission data: The first related sequence is set according to relevant factors (such as total population X2, per capita GDP X3, energy consumption X4, and total value of secondary industry X5):

[0052]

[0053] Among them, l, m, and n are positive integers and l+m>n, which indicates that the length of the correlation factor sequence (ie, the first correlation sequence) is longer than the length of the system behavior sequence.

[0054] After obtaining the system behavior sequence and multiple first correlation sequences, it is necessary to determine the factors that are strongly correlated with carbon emissions among the multiple correlation factors corresponding to the first correlation sequences. The specific steps are as follows.

[0055] In some embodiments of the present application, the length of the first correlation sequence is greater than the length of the system behavior sequence; determining the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as the second correlation sequence includes the following steps: generating a time-lag sequence corresponding to the first correlation sequence, wherein the length of each time-lag sequence is the same as the length of the system behavior sequence, and each time-lag sequence corresponds to a lag period, and the lag period is used to characterize the lag between the first correlation sequence corresponding to the time-lag sequence and the system behavior sequence in the time dimension; determining the correlation parameter based on the time-lag sequence and the system behavior sequence; when the lag period corresponding to the time-lag sequence characterizes that the first correlation sequence lags behind the system behavior sequence, and the correlation parameter is greater than the preset correlation threshold, determining the first correlation sequence corresponding to the time-lag sequence as the second correlation sequence.

[0056] Specifically, the lag sequence corresponding to the correlation factor (first correlation sequence) is assumed to be: Among them, p i is the lag period; calculate the p corresponding to the relevant factor (first relevant sequence) i Period-lag series and system behavior series Grey absolute correlation (i.e. correlation parameter) ε i,p , as shown in the following formula:

[0057]

[0058] in,

[0059] Then, the gray absolute correlation (i.e. correlation parameter) ε i,p Compared with the preset correlation threshold γ, if p i >0 and ε i,p ≥γ, it means that the sequence can be used as a driving factor, that is, the first correlation sequence whose correlation parameter is greater than the preset correlation threshold is determined as the second correlation sequence. i <0 and ε i,p ≥γ, it means that the sequence does not meet the modeling conditions. For example, in this embodiment, the total population X2, the total consumption of the three major energy sources X4, and the output value of the secondary industry X5 are determined as strong correlation factors.

[0060] In addition, after determining the strong correlation factors (the second correlation sequence), a time-lag grey model (GM (1, N)) can also be constructed. The specific steps are as follows.

[0061] In some embodiments of the present application, the method also includes the following steps: determining a first-order cumulative generation sequence corresponding to a second related sequence, wherein the kth element in the first-order cumulative generation sequence corresponding to the second related sequence is the cumulative sum of the 1st to kth elements in the second related sequence, and k is a positive integer; determining a neighboring mean generation sequence corresponding to a system behavior sequence, wherein the kth element in the neighboring mean generation sequence is the mean of the kth element and the k-1th element in the first-order cumulative generation sequence corresponding to the system behavior sequence; determining a time-lag gray model based on the first-order cumulative generation sequence corresponding to the second related sequence, the neighboring mean generation sequence corresponding to the system behavior sequence, and the lag period corresponding to the second related sequence, wherein the time-lag gray model is used to characterize the correlation between the carbon emission-related factors represented by the second related sequence and the carbon emissions.

[0062] Specifically, a first-order cumulative generation (1-AGO) sequence is generated And generate The corresponding adjacent mean generation sequence in, Generates the first-order cumulative sequence corresponding to the system behavior sequence.

[0063] Then, combining the relevant factors and the lag period p i , establish the time-delay GM(1,N) model, the model formula is as follows:

[0064]

[0065] Among them, a is the system development coefficient, is the driving term, b i is the driving coefficient, p i is the lag parameter (lag period).

[0066] After determining the strong correlation factor (second correlation sequence), univariate prediction can also be performed through a univariate probabilistic time series model (such as the Lag-Llama time series prediction model), that is, based on the second correlation sequence, determine the third correlation sequence corresponding to the target time period. The specific steps are as follows.

[0067] In some embodiments of the present application, determining a third related sequence corresponding to a target time period based on a second related sequence includes the following steps: determining a feature sequence corresponding to the second related sequence, and performing standardization on the feature sequence; encoding the position information of each element in the standardized feature sequence through sine and cosine position coding to obtain a position code; mapping the feature sequence to a decoder of a univariate probability time series model based on the position code, and using the decoder to perform autoregressive prediction based on the feature sequence to obtain a third related sequence, wherein a t distribution head is used in the univariate probability time series model to determine the probability distribution of the result output by the decoder.

[0068] Specifically, the present embodiment considers the case of a single variable. Different data sets usually contain different variable data. Instead of sampling or filling different data sets to a specified dimension, it is better to start directly from a single sequence. Single variable prediction is simpler and can be better generalized to different data sets. Lag-Llama is built on the Llama architecture of Transformer, such as Figure 5 As shown in the figure, it contains M decoding layers, which map features to the hidden dimensions of the attention module through the univariate sequence and the covariate labeled shared linear projection layer. Pre-normalization is achieved through RMS Norm, and Rotation Position Encoding (RoPE) is added to the attention layer to optimize the prediction performance of the model by minimizing the negative log-likelihood on the prediction window. Trained on a large corpus of time series data, the model can generalize unseen time series well and perform zero-sample prediction, which is described in detail below.

[0069] First, construct the feature sequence input and summary statistical information corresponding to the second related sequence. The specific steps are as follows.

[0070] In some embodiments of the present application, standardization of a feature sequence includes the following steps: determining the mean and variance of the elements in the feature sequence, and standardizing each element in the feature sequence based on the mean and variance; determining the root mean square corresponding to the elements in the feature sequence after standardization, and normalizing each element in the feature sequence after standardization based on the root mean square, wherein the normalization is used to map each element to the same scale range.

[0071] Specifically, the feature sequence can be expressed as: Calculate the mean: variance: Each element in the feature sequence is standardized according to the mean and variance to obtain the final summary statistical information representation: That is, after the feature sequence is standardized, the data can be converted to a unified scale. In this embodiment, the RMS Norm method can be used, as follows:

[0072]

[0073] Among them, RMS(x) represents the root mean square corresponding to the elements in the feature sequence, and d represents the total dimension of the position encoding vector.

[0074] After that, the position information of each element in the standardized feature sequence is encoded by sine-cosine position coding. Specifically, in this embodiment, the position encoder adopts an absolute position encoder, and the position information of each token is encoded by using Sinusoidal-sine-cosine position coding. The Sinusoidal position coding expression formula is as follows:

[0075]

[0076] Among them, pos represents the position in the sequence, i represents the dimension index of the position encoding vector, and d represents the total dimension of the position encoding vector.

[0077] Then, the feature sequence is mapped to the decoder module through position coding. In this embodiment, the t distribution head is used to determine the probability distribution of the result output by the decoder. Specifically, the student t distribution can be used, because the student t distribution can be regarded as an extension of the normal distribution. In the case of small samples or high uncertainty, the student t distribution can provide better theoretical support and explanation. The probability distribution is calculated by the feature sequence. The probability density function of the student t distribution can be expressed as:

[0078]

[0079] where υ is the degree of freedom and Γ is the gamma function:

[0080] The feature sequence passes through autoregression, and the model will generate a prediction sequence of set length, which is the third related sequence mentioned above.

[0081] After obtaining the predicted value corresponding to the strong correlation factor determined by the Lag-Llama model (i.e., the third correlation sequence), the time-lag gray model can be used to determine the carbon emission data corresponding to the target time period based on the third correlation sequence. Specifically, the predicted value of the single variable (i.e., the third correlation sequence) can be brought into the time-lag characteristic GM(1,N) formula to calculate the final carbon emission prediction value.

[0082] The following is an example of the method for determining the carbon emissions in steps S202 to S208.

[0083] The historical carbon emission data and the historical data of carbon emission related factors of a province are obtained. In this embodiment, the statistical time is unified from 1997 to 2021. When calculating the time-lag GM (1, N), due to the order of magnitude difference between the data, the data is converted to a unified scale, and the relevant driving factors are determined based on the time-lag GM (1, N). The correlation parameters corresponding to each related factor (first related sequence) are shown in the following table.

[0084] <![CDATA[X2]]> <![CDATA[X3]]> <![CDATA[X4]]> <![CDATA[X5]]> 0.8816 0.7659 0.9491 0.9009

[0085] From the results, it can be seen that if the correlation threshold is selected as 0.85, the strongly correlated factors (i.e., the second correlation sequence) include: total population X2, total consumption of the three major energy sources X4, and gross output value of the secondary industry X5; and per capita GDP X3 is not a strongly correlated factor in this embodiment.

[0086] After that, Lag-Llama can be used to predict the total population, total consumption of the three major energy sources, and the gross output value of the secondary industry of the province from 2021 to 208, as follows. The second related series corresponding to the total population, total consumption of the three major energy sources, and gross output value of the secondary industry are converted into specific time series, and characteristic series are constructed to quantify the data on a unified scale. The quantification rule used in this embodiment is RMS Norm; the data in the statistical yearbook are selected from consecutive years, and the data from 1978 to 2021 are normalized as follows:

[0087] years <![CDATA[X2 - Population]]> <![CDATA[X4 - Energy]]> <![CDATA[X5-Industry]]> 1978 -1.15 -0.726 -0.5319 1979 -1.1 -0.72 -0.5319 1980 -1.05 -0.723 -0.5317 ... ... ... ... 2020 1.24 0.754 1.254 2021 1.27 0.917 1.507

[0088] Initialization model: predict the population data based on Lag-Llama's zero-sample prediction model, set the prediction length to 7, and the text length to 44; perform model prediction. In this embodiment, the final prediction results for the total population, the total consumption of the three major energies, and the total value of the secondary industry are shown in the following table:

[0089] years <![CDATA[X2]]> <![CDATA[X4]]> <![CDATA[X5]]> 2022 4188 15659.05 369368.5 2023 4227 16322.95 394252.5 2024 4290 16800.42 405199.8 2025 4310 17204.52 419326.5 2026 4349 17796.91 436183.9 2027 4386 18071.94 445947.1 2028 4401 18507.87 455935.0

[0090] Since the time dimension statistics of carbon-related data are not uniform and the data volume is insufficient, the use of zero-sample prediction can achieve effective results and the text solution can correct the real-time data.

[0091] Afterwards, the predicted values ​​of X2, X4, and X5 are brought into the above GM(1,N) model to calculate the final carbon emissions forecast as shown in the following table.

[0092] 2022 2023 2024 2025 2026 2027 2028 325.82 344.10 351.1 357.83 372.24 377.91 382.09

[0093] Due to the complexity of the carbon emission system, policies and effects have a certain time lag. Therefore, the GM(1,N) model with time lag characteristics is used in the prediction method. Combining multimodal data such as urban expansion and other factors that are closely related to carbon emissions, the accuracy of calculating carbon emissions has been improved. Figure 6 shown.

[0094] The present application adopts the Lag-Llama univariate time series prediction model to predict factors strongly associated with carbon emissions, so as to improve the accuracy of the entire prediction process and reduce the interference of other factors. Compared with traditional time series prediction, the generalization ability and calculation accuracy of the model are improved. In the calculation of carbon emissions, the GM (1, N) model with time lag characteristics is used to find the optimal carbon emission mode and predict carbon emissions, which ensures the accuracy of the model in small sample conditions and improves the generalization ability of the model. At the same time, the time lag effect caused by policies over a long period of time can be considered to a greater extent. Due to the complexity of the carbon emission system, combined with the Lag-Llama univariate prediction, the mutual interference of other factors in the calculation process can be excluded, making the predicted value more valuable for reference and improving the reusability. The carbon emission calculation method of the present application mainly starts from the factors related to carbon emissions, simulates the impact of other factors on carbon emissions, and thus constructs a related model framework. Based on the Lag-Llama univariate prediction, it can also better avoid the redundant calculation error superposition caused by the same algorithm.

[0095] According to an embodiment of the present application, an embodiment of a device for determining carbon emissions is also provided. Figure 7 Schematic diagram of a carbon emission determination device provided according to an embodiment of the present application. Figure 7 As shown, the device comprises:

[0096] The data sequence acquisition module 70 is used to generate a system behavior sequence and a first related sequence based on the historical carbon emission data and the historical data of carbon emission related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first related sequence is a time series of historical values ​​of factors related to carbon emissions;

[0097] A strong correlation factor determination module 72 is used to determine the correlation parameter between the system behavior sequence and each first correlation sequence, and determine the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as a second correlation sequence, wherein the correlation parameter is used to characterize the influence of the carbon emission related factors on the carbon emissions;

[0098] The univariate prediction module 74 is used to determine a third correlation sequence corresponding to the target time period based on the second correlation sequence, wherein the third correlation sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second correlation sequence;

[0099] The carbon emission determination module 76 is used to determine the carbon emission data corresponding to the target time period according to the third related sequence.

[0100] Optionally, the length of the first correlation sequence is greater than the length of the system behavior sequence; determining the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as the second correlation sequence includes: generating a time-lag sequence corresponding to the first correlation sequence, wherein the length of each time-lag sequence is the same as the length of the system behavior sequence, and each time-lag sequence corresponds to a lag period, and the lag period is used to characterize the lag between the first correlation sequence corresponding to the time-lag sequence and the system behavior sequence in the time dimension; determining the correlation parameter based on the time-lag sequence and the system behavior sequence; when the lag period corresponding to the time-lag sequence characterizes that the first correlation sequence lags behind the system behavior sequence, and the correlation parameter is greater than the preset correlation threshold, determining the first correlation sequence corresponding to the time-lag sequence as the second correlation sequence.

[0101] Optionally, determining the third related sequence corresponding to the target time period based on the second related sequence includes: determining a feature sequence corresponding to the second related sequence, and standardizing the feature sequence; encoding the position information of each element in the standardized feature sequence through sine and cosine position coding to obtain a position code; based on the position code, mapping the feature sequence to a decoder of a univariate probability time series model, and using the decoder to perform autoregressive prediction based on the feature sequence to obtain a third related sequence, wherein a t distribution head is used in the univariate probability time series model to determine the probability distribution of the result output by the decoder.

[0102] Optionally, standardizing the feature sequence includes: determining the mean and variance of the elements in the feature sequence, and standardizing each element in the feature sequence based on the mean and variance; determining the root mean square corresponding to the elements in the feature sequence after standardization, and normalizing each element in the feature sequence after standardization based on the root mean square, wherein the normalization is used to map each element to the same scale range.

[0103] Optionally, the carbon emission determination device is also used to: determine a first-order cumulative generation sequence corresponding to the second related sequence, wherein the kth element in the first-order cumulative generation sequence corresponding to the second related sequence is the cumulative sum of the 1st to kth elements in the second related sequence, and k is a positive integer; determine a neighboring mean generation sequence corresponding to the system behavior sequence, wherein the kth element in the neighboring mean generation sequence is the mean of the kth element and the k-1th element in the first-order cumulative generation sequence corresponding to the system behavior sequence; determine a time-lag gray model based on the first-order cumulative generation sequence corresponding to the second related sequence, the neighboring mean generation sequence corresponding to the system behavior sequence, and the lag period corresponding to the second related sequence, wherein the time-lag gray model is used to characterize the correlation between the carbon emission-related factors represented by the second related sequence and the carbon emissions.

[0104] Optionally, determining the carbon emission data corresponding to the target time period according to the third related sequence includes: using a time-lag grey model to determine the carbon emission data corresponding to the target time period according to the third related sequence.

[0105] Optionally, generating a system behavior sequence and a first related sequence based on historical carbon emission data and historical data of carbon emission related factors includes: acquiring historical carbon emission data and historical data of carbon emission related factors, wherein the historical carbon emission data includes at least one of the following: carbon emission data of previous years, and the historical data of carbon emission related factors includes at least one of the following: total population of previous years, per capita GDP, total consumption of coal, natural gas and oil, and total value of secondary industry; generating a system behavior sequence based on carbon emission data of previous years; generating a first related sequence based on historical data of carbon emission related factors, wherein each carbon emission related factor corresponds to a first related sequence.

[0106] It should be noted that the various modules in the above-mentioned carbon emission determination device can be program modules (for example, a set of program instructions that implement a certain specific function) or hardware modules. For the latter, it can be expressed in the following forms, but is not limited to this: the expression form of each of the above-mentioned modules is a processor, or the functions of each of the above-mentioned modules are implemented by a processor.

[0107] It should be noted that the carbon emission determination device provided in this embodiment can be used to perform Figure 2 The carbon emission determination method shown, therefore, the relevant explanations and instructions on the above-mentioned carbon emission determination method are also applicable to the embodiments of the present application and will not be repeated here.

[0108] An embodiment of the present application also provides a non-volatile storage medium, the non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the following carbon emissions determination method by running the computer program: based on historical carbon emissions data and historical data of carbon emission-related factors, a system behavior sequence and a first related sequence are generated, wherein the system behavior sequence is a time series of historical carbon emissions, and the first related sequence is a time series of historical values ​​of factors related to carbon emissions; correlation parameters between the system behavior sequence and each first related sequence are determined, and the first related sequence whose correlation parameter is greater than a preset correlation threshold is determined as a second related sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; based on the second related sequence, a third related sequence corresponding to a target time period is determined, wherein the third related sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period located after the historical time period corresponding to the second related sequence; based on the third related sequence, the carbon emissions data corresponding to the target time period is determined.

[0109] The embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the carbon emissions determination method described in each embodiment of the present application: generating a system behavior sequence and a first related sequence based on historical carbon emissions data and historical data of carbon emission-related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first related sequence is a time series of historical values ​​of factors related to carbon emissions; determining a correlation parameter between the system behavior sequence and each first related sequence, and determining a first related sequence whose correlation parameter is greater than a preset correlation threshold as a second related sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; determining a third related sequence corresponding to a target time period based on the second related sequence, wherein the third related sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period located after the historical time period corresponding to the second related sequence; determining carbon emissions data corresponding to the target time period based on the third related sequence.

[0110] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0111] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0112] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0113] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0114] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0116] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for determining carbon emissions, characterized in that: include: Generate a system behavior sequence and a first correlation sequence based on historical carbon emission data and historical data of carbon emission related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first correlation sequence is a time series of historical values ​​of factors related to carbon emissions; Determine a correlation parameter between the system behavior sequence and each of the first correlation sequences, and determine the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as a second correlation sequence, wherein the correlation parameter is used to characterize the degree of influence of carbon emission-related factors on carbon emissions; Determine a third related sequence corresponding to a target time period according to the second related sequence, wherein the third related sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second related sequence; The carbon emission data corresponding to the target time period is determined according to the third related sequence.

2. The method for determining carbon emissions according to claim 1, characterized in that: The length of the first correlation sequence is greater than the length of the system behavior sequence; Determining the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as a second correlation sequence includes: Generate a time lag sequence corresponding to the first correlation sequence, wherein the length of each of the time lag sequences is the same as the length of the system behavior sequence, and each of the time lag sequences corresponds to a lag period, and the lag period is used to characterize the lag between the first correlation sequence corresponding to the time lag sequence and the system behavior sequence in the time dimension; Determining the correlation parameter according to the time delay sequence and the system behavior sequence; When the lag period corresponding to the time lag sequence represents that the first correlation sequence lags behind the system behavior sequence, and when the correlation parameter is greater than a preset correlation threshold, the first correlation sequence corresponding to the time lag sequence is determined as the second correlation sequence.

3. The method for determining carbon emissions according to claim 2, characterized in that: Determining, according to the second correlation sequence, a third correlation sequence corresponding to the target time period includes: Determining a characteristic sequence corresponding to the second correlation sequence, and performing standardization processing on the characteristic sequence; By means of sine and cosine position coding, the position information of each element in the characteristic sequence after the standardization process is encoded to obtain a position code; According to the position coding, the feature sequence is mapped to a decoder of a univariate probability time series model, and the decoder is used to perform autoregressive prediction based on the feature sequence to obtain the third related sequence, wherein a t distribution head is used in the univariate probability time series model to determine the probability distribution of the result output by the decoder.

4. The method for determining carbon emissions according to claim 3, characterized in that: The standardization process for the feature sequence includes: Determine the mean and variance of the elements in the feature sequence, and perform standardization processing on each element in the feature sequence according to the mean and the variance; Determine the root mean square corresponding to the elements in the feature sequence after the standardization processing, and perform normalization processing on each element in the feature sequence after the standardization processing according to the root mean square, wherein the normalization processing is used to map each of the elements to the same scale range.

5. The method for determining carbon emissions according to claim 2, characterized in that: The method further comprises: Determine a first-order cumulative generation sequence corresponding to the second correlation sequence, wherein a kth element in the first-order cumulative generation sequence corresponding to the second correlation sequence is a cumulative sum of elements 1 to k in the second correlation sequence, and k is a positive integer; Determine a neighboring mean generation sequence corresponding to the system behavior sequence, wherein the kth element in the neighboring mean generation sequence is the mean of the kth element and the k-1th element in the first-order accumulation generation sequence corresponding to the system behavior sequence; A time-lag grey model is determined based on the first-order cumulative generation sequence corresponding to the second correlation sequence, the adjacent mean generation sequence corresponding to the system behavior sequence, and the lag period corresponding to the second correlation sequence, wherein the time-lag grey model is used to characterize the correlation between the carbon emission-related factors represented by the second correlation sequence and the carbon emissions.

6. The method for determining carbon emissions according to claim 5, characterized in that: Determining the carbon emission data corresponding to the target time period according to the third correlation sequence includes: The time-lag grey model is adopted to determine the carbon emission data corresponding to the target time period according to the third related sequence.

7. The method for determining carbon emissions according to claim 1, characterized in that: Based on historical carbon emission data and historical data of carbon emission related factors, the system behavior sequence and the first related sequence are generated, including: Acquire historical carbon emission data and historical data of carbon emission related factors, wherein the historical carbon emission data includes at least one of the following: carbon emission data of previous years, and the historical data of carbon emission related factors includes at least one of the following: total population, per capita GDP, total coal, natural gas and oil consumption, and total value of secondary industry of previous years; Generate the system behavior sequence according to the carbon emission data of previous years; The first correlation sequence is generated according to the historical data of the carbon emission related factors, wherein each of the carbon emission related factors corresponds to one first correlation sequence.

8. A device for determining carbon emissions, characterized in that: include: A data sequence acquisition module, used to generate a system behavior sequence and a first related sequence based on historical carbon emission data and historical data of carbon emission related factors, wherein the system behavior sequence is a time series of historical carbon emissions, and the first related sequence is a time series of historical values ​​of factors related to carbon emissions; A strong correlation factor determination module, used to determine the correlation parameter between the system behavior sequence and each of the first correlation sequences, and determine the first correlation sequence whose correlation parameter is greater than a preset correlation threshold as a second correlation sequence, wherein the correlation parameter is used to characterize the influence of the carbon emission related factors on the carbon emissions; a univariate prediction module, configured to determine a third correlation sequence corresponding to a target time period based on the second correlation sequence, wherein the third correlation sequence is a sequence of values ​​of factors related to carbon emissions arranged in chronological order within the target time period, and the target time period is a time period after the historical time period corresponding to the second correlation sequence; The carbon emission determination module is used to determine the carbon emission data corresponding to the target time period according to the third related sequence.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program executes the carbon emission determination method according to any one of claims 1 to 7 when running.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the carbon emission determination method according to any one of claims 1 to 7 by running the computer program.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the carbon emission determination method according to any one of claims 1 to 7 are implemented.

Citation Information

Cited By

  • Time sequence prediction model generation method, data processing method and device

    CN119272821A