Carbon asset evaluation method, system and equipment based on online prediction and medium
Through the combination of online prediction methods and neural network models, the carbon asset appraisal model is dynamically corrected, which solves the problems of low efficiency and low accuracy of traditional evaluation methods, and achieves efficient and accurate assessment of carbon assets, supporting the realization of low-carbon goals in the park. .
Patent Information
- Application Number
- CN202410515576.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-26
- Publication Date
- 2024-11-08
AI Technical Summary
The existing carbon asset evaluation methods are inefficient and have low accuracy. Relying on human resources assessment and traditional neural network models, it is difficult to accurately evaluate the carbon emission levels in the park in real time.
Using an online prediction method, by acquiring and preprocessing carbon asset information data, a long and short-term memory neural network (LSTM) and gate recurrent unit neural network (GRU) model is constructed, and offline training and online incremental update are combined with hedging algorithms. Correct the evaluation model to improve evaluation accuracy.
It has achieved rapid identification and efficient correction of the changes in carbon asset information data and improved the accuracy and real-time nature of carbon emission assessment, and helped the park optimize management and achieve low-carbon goals.
Smart Images

Figure CN118917449A_ABST
Abstract
Description
Technical Field
[0001] This document relates to the technical field of carbon asset assessment, and in particular to a carbon asset assessment method, system, equipment and medium based on online prediction. Background Art
[0002] As the core unit of industrial cluster development, the industrial park has abundant carbon assets. Carbon assets have become an important element in the development and competition of various industrial parks. It is particularly important for the economic development of the industrial park to understand the trend of low-carbon development and effectively manage and enhance the value of carbon assets.
[0003] At present, the rating method of carbon assets is mainly through evaluation institutions or personnel, which is not only labor-intensive but also inefficient. The commonly used automated evaluation method mainly uses neural network models for evaluation. Although this method has certain results, it is not accurate and has low real-time performance. Therefore, it is necessary to provide an efficient method for evaluating carbon assets. Summary of the invention
[0004] The present invention provides a carbon asset assessment method, system, device and medium based on online prediction, aiming to solve the above problems.
[0005] The embodiment of the present invention provides a carbon asset assessment method based on online prediction, comprising:
[0006] Obtain carbon asset information data, and divide the carbon asset information data into carbon asset historical data and carbon asset latest data;
[0007] Pre-process the historical data and the latest data of carbon assets;
[0008] The carbon asset benchmark assessment model is obtained through an offline training method to perform carbon asset assessment and obtain assessment data. According to the latest pre-processed carbon asset data, it is determined whether the carbon asset benchmark assessment model should be revised. If revision is required, the carbon asset assessment is performed through the revised carbon asset benchmark assessment model to obtain assessment data.
[0009] The embodiment of the present invention provides a carbon asset evaluation system based on online prediction, comprising:
[0010] A data acquisition module is used to acquire carbon asset information data and divide the carbon asset information data into carbon asset historical data and carbon asset latest data;
[0011] Data processing module, used to pre-process historical data and latest data of carbon assets;
[0012] Carbon asset assessment module, including training submodule, change identification submodule and online increment submodule;
[0013] The training submodule is used to obtain a carbon asset benchmark assessment model through an offline training method to perform carbon asset assessment and obtain assessment data;
[0014] The change identification submodule is used to obtain the change range between the latest data of pre-processed carbon assets and the assessment data obtained by the carbon asset benchmark assessment model;
[0015] The online incremental submodule is used to determine whether to revise the carbon asset benchmark assessment model based on the latest pre-processed carbon asset data. If revision is required, the carbon asset assessment is performed using the revised carbon asset benchmark assessment model to obtain assessment data.
[0016] An embodiment of the present invention further provides an electronic device, including:
[0017] processor; and,
[0018] A memory arranged to store computer executable instructions, which, when executed, cause the processor to perform the steps of the above-mentioned carbon asset assessment method based on online prediction.
[0019] An embodiment of the present invention further provides a storage medium for storing computer executable instructions, which, when executed, implement the steps of the above-mentioned carbon asset assessment method based on online prediction.
[0020] By adopting the embodiment of the present invention, when the carbon asset information data changes, it can be quickly identified and whether the benchmark assessment model needs to be revised is determined according to the magnitude of the change. According to the result of the change identification, and according to the magnitude of the change in the carbon asset information data, it is determined whether the carbon asset benchmark assessment model needs to be revised. Compared with the traditional offline assessment method, it has higher accuracy and can more accurately assess the carbon emission level of the park, help the park discover problems, optimize management, and promote the realization of low-carbon goals. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0022] Figure 1 A flowchart of a carbon asset assessment method based on online prediction according to an embodiment of the present invention;
[0023] Figure 2 A schematic diagram of a carbon asset assessment system based on online prediction according to an embodiment of the present invention;
[0024] Figure 3 This is a schematic diagram of an embodiment of the present invention using a carbon asset benchmark assessment model to perform carbon asset assessment. DETAILED DESCRIPTION
[0025] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.
[0026] Method Embodiment
[0027] According to an embodiment of the present invention, a carbon asset evaluation method based on online prediction is provided. Figure 1 is a flow chart of a carbon asset assessment method based on online prediction according to an embodiment of the present invention. Figure 1 As shown, the carbon asset assessment method based on online prediction in the embodiment of the present invention specifically includes:
[0028] S1. Obtain carbon asset information data, and divide the carbon asset information data into carbon asset historical data and carbon asset latest data;
[0029] In step S1, a carbon asset assessment service request sent by a user terminal is received, carbon asset information data is received, and the carbon asset information data is divided into two categories: historical data and latest data. The carbon asset historical data is used for the subsequent offline training of the carbon asset benchmark assessment model. Among them, the carbon asset information data includes: the carbon emissions of the park, the quota of the park's carbon assets, the average transaction price of the market's effective carbon quota, the intensity of energy conservation and emission reduction, etc.
[0030] S2. Preprocessing the historical data of carbon assets and the latest data of carbon assets;
[0031] The method for preprocessing the carbon asset information data in step S2 is de-averaging and normalization. The preprocessed data has higher validity than the original data.
[0032] S3. Obtain a carbon asset benchmark assessment model through an offline training method to perform carbon asset assessment and obtain assessment data. Determine whether to revise the carbon asset benchmark assessment model based on the latest preprocessed carbon asset data. If revision is required, perform carbon asset assessment and obtain assessment data through the revised carbon asset benchmark assessment model.
[0033] like Figure 3 As shown, step S3 can be further divided into three steps:
[0034] S31, offline training part;
[0035] S32, change identification part;
[0036] S33, online incremental update part;
[0037] In the offline training part S31, the historical data of carbon assets obtained in S1 are used for offline training to obtain the carbon asset benchmark assessment model. First, the pre-processed historical data of carbon assets are divided into a training set and a validation set according to a specific ratio. The ratio of the training set to the validation set used in the embodiment of the present invention is 7:3. The specific process of offline training is as follows:
[0038] The long short-term memory neural network (LSTM) and gated recurrent unit neural network (GRU) are constructed respectively. The training set is trained through the offline training part, and then verified with the verification set. There are input gates, forget gates and output gates inside the LSTM. The forget gate is used to store the weight vector of the old information, the input gate is used to obtain the weight vector of the new information, and the output gate is used to output the candidate vector to be output. The LSTM expression specifically includes:
[0039]
[0040] Among them, X t represents the input at time t, h t represents the output at time t, h t-1 represents the output at time t-1, C t and C t-1 are the cell states at time t and t-1 respectively, is the unit state transition value, f t 、i t and O t They represent the forget gate, input gate and output gate respectively, σ is the sigmoid activation function, tanh is the tanh activation function, W fh , W ih , W oh , W ch Represents the interconnection weight matrix of forget gate, input gate, output gate, and unit state, W fx , W ix , W ox , W cx are the input weight matrices of the forget gate, input gate, output gate, and unit state, respectively, and b f 、b i 、b o 、b c They are the corresponding bias items for the forget gate, input gate, output gate, and unit state output respectively. Represents the Hadamard product of matrices.
[0041] GRU has update gate and recharge gate. Compared with LSTM, it is more concise in structure. The specific expression is as follows:
[0042]
[0043] Among them, z t represents the update gate input at time t, r t represents the reset gate input at time t, is the transition value, W zx , W rx , W zh , W rh Represent the weight matrices of the update gate and reset gate, W h The weight matrix representing the transition value, b z 、b r Output the corresponding bias terms for the update gate and reset gate respectively.
[0044] The S32 part quickly identifies changes in carbon asset information data, and determines whether the constructed benchmark assessment model needs to be revised based on the magnitude of the change. We first set a carbon asset change threshold. If the data change magnitude is less than the carbon asset change threshold, the carbon asset benchmark assessment model of the previous moment will continue to be used. If the data change magnitude is greater than the carbon asset change threshold, the assessment model will be revised to obtain a new carbon asset assessment model. Changes in carbon asset information data are mainly short-term changes, such as fluctuations in the average transaction price of effective carbon quotas in the market, changes in carbon emissions due to emergencies, etc. According to the results of change identification, when the change amount is greater than the threshold, the online incremental update part in S33 is enabled.
[0045] In S33, online incremental updates are performed based on the hedging algorithm, and a dual-channel neural network based on the hedging algorithm is designed. The dual-channel neural network is formed by the two independent paths of LSTM and GRU mentioned above. The concept of hedging algorithm is introduced on this basis, and output weights are assigned to the outputs of LSTM and GRU and adjusted in real time. When the LSTM accuracy is higher, the output weight corresponding to LSTM is increased, and when the GRU accuracy is higher, the output weight corresponding to GRU is increased. When the carbon asset information data changes, if the change is greater than the threshold, the output weights of LSTM and GRU are adjusted, and the overall model is updated by backpropagation. If the change is less than the threshold, the overall model is not updated. The hedging algorithm is as follows:
[0046] First, each module in the LSTM and GRU channels performs forward propagation, and the expression is:
[0047]
[0048] in, Represents the initial values of LSTM and GRU Represents the input value of the lth layer of LSTM and GRU, Represents the output value of the lth layer of LSTM and GRU, FC refers to the fully connected layer used for residual connection, concat represents vector cascade calculation, l∈(0,1,2…L), L is the number of module layers for each channel.
[0049] Then, the output of each module is calculated through the fully connected layer to obtain the prediction result of each module, and then these results are weighted and summed to form the final prediction output of the model:
[0050]
[0051] in, It represents the prediction result and output weight of the lth layer of LSTM and GRU, and yy represents the final output.
[0052] Finally, when the detected error is greater than the threshold, the output weight is adjusted And back-propagate to update the overall model.
[0053] By adopting the embodiments of the present invention, the following beneficial effects are achieved:
[0054] When the carbon asset information data changes, it can be quickly identified and whether the benchmark assessment model needs to be revised based on the magnitude of the change. Based on the results of the change identification, it can determine whether the carbon asset benchmark assessment model needs to be revised based on the magnitude of the change in the carbon asset information data. Compared with traditional offline assessment methods, it has higher accuracy and can more accurately assess the carbon emission level of the park, help the park identify problems, optimize management, and promote the realization of low-carbon goals.
[0055] System Example
[0056] The embodiment of the present invention provides a carbon asset evaluation system based on online prediction. Figure 2 A schematic diagram of an online prediction carbon asset evaluation system according to an embodiment of the present invention. Figure 2 As shown, the carbon asset assessment system based on online prediction in the embodiment of the present invention specifically includes:
[0057] The data acquisition module 20 is used to acquire carbon asset information data and divide the carbon asset information data into historical carbon asset data and the latest carbon asset data; the data acquisition module 20 receives the carbon asset assessment service request sent by the user terminal, receives the carbon asset information data, and divides the carbon asset information data into historical data and the latest data. The historical carbon asset data is used for the offline training of the carbon asset benchmark assessment model. Among them, the carbon asset information data includes: the carbon emissions of the park, the quota of the park's carbon assets, the average transaction price of the market's effective carbon quota, the intensity of energy conservation and emission reduction, etc.
[0058] The data processing module 22 is used to pre-process the historical data and the latest data of carbon assets. The method of pre-processing the carbon asset information data in the data processing module 22 is de-averaging and normalization. The pre-processed data has a higher validity than the original data.
[0059] A carbon asset assessment module 24, including a training submodule, a change identification submodule and an online increment submodule;
[0060] The training submodule is used to obtain a carbon asset benchmark assessment model through an offline training method to perform carbon asset assessment and obtain assessment data;
[0061] The change identification submodule is used to obtain the change range between the latest data of pre-processed carbon assets and the assessment data obtained by the carbon asset benchmark assessment model;
[0062] The online incremental submodule is used to determine whether to revise the carbon asset benchmark assessment model based on the latest pre-processed carbon asset data. If revision is required, the carbon asset assessment is performed using the revised carbon asset benchmark assessment model to obtain assessment data.
[0063] Among them, the training submodule is specifically used for:
[0064] A carbon asset benchmark assessment model is constructed based on the long short-term memory neural network LSTM and the recurrent unit neural network GRU. The historical data of carbon assets are divided into a training set and a validation set. The training set is trained through the offline training part and verified with the validation set.
[0065] An input gate, a forget gate, and an output gate are set in the long short-term memory neural network LSTM. The forget gate is used to store the weight vector of old information, the input gate is used to obtain the weight vector of new information, and the output gate is used to output the candidate vector to be output.
[0066] The expression of long short-term memory neural network LSTM specifically includes:
[0067]
[0068] Among them, X t represents the input at time t, ht represents the output at time t, h t-1 represents the output at time t-1, C t and C t-1 are the cell states at time t and t-1 respectively, is the unit state transition value, f t 、i t and O t They represent the forget gate, input gate and output gate respectively, σ is the sigmoid activation function, tanh is the tanh activation function, W fh , W ih , W oh , W ch Represents the interconnection weight matrix of forget gate, input gate, output gate, and unit state, W fx , W ix , W ox , W cx are the input weight matrices of the forget gate, input gate, output gate, and unit state, respectively, and b f 、b i 、b o 、b c They are the corresponding bias items for the forget gate, input gate, output gate, and unit state output respectively. Represents the Hadamard product of matrices.
[0069] The recurrent unit neural network GRU includes an update gate and a reset gate, and the recurrent unit neural network is obtained by formula 2;
[0070]
[0071] Among them, z t represents the update gate input at time t, r t represents the reset gate input at time t, is the transition value, W zx , W rx , W zh , W rh Represent the weight matrices of the update gate and reset gate, W h The weight matrix representing the transition value, b z 、b r Output the corresponding bias terms for the update gate and reset gate respectively.
[0072] The online incremental submodule is specifically used for:
[0073] When the change in the latest data of carbon assets after preprocessing is greater than a preset threshold compared with the assessment data obtained by the carbon asset benchmark assessment model, the hedging algorithm is used to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network, and backpropagation is used to update the carbon asset benchmark assessment model.
[0074] The online incremental submodule uses a hedging algorithm to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network and back-propagate to update the carbon asset benchmark assessment model. Specifically, it includes:
[0075] Forward propagation is performed on each module in the long short-term memory neural network and the recurrent unit neural network, and the expression is:
[0076]
[0077] in, represents the initial value of the short-term memory neural network and the recurrent unit neural network, represents the input value of the lth layer of the short-term memory neural network and the recurrent unit neural network, Represents the output value of the lth layer of the short-term memory neural network and the recurrent unit neural network, FC is the fully connected layer used for residual connection, concat represents vector cascade calculation, l∈(0,1,2…L), L is the number of module layers for each channel;
[0078] The output of each module is calculated through the fully connected layer to obtain the prediction result of each module, and the prediction results are weighted and summed to form the final prediction output of the model:
[0079]
[0080] in, represents the prediction results and output weights of the lth layer of the long short-term memory neural network, represents the prediction result and output weight of the lth layer of the recurrent unit neural network, and yy represents the final output;
[0081] When the error of the final output is greater than the preset threshold, the output weight is adjusted And back-propagate to update the overall model.
[0082] The embodiments of the present invention have the following beneficial effects:
[0083] When the carbon asset information data changes, it can be quickly identified and whether the benchmark assessment model needs to be revised based on the magnitude of the change. Based on the results of the change identification, it can determine whether the carbon asset benchmark assessment model needs to be revised based on the magnitude of the change in the carbon asset information data. Compared with traditional offline assessment methods, it has higher accuracy and can more accurately assess the carbon emission level of the park, help the park identify problems, optimize management, and promote the realization of low-carbon goals.
[0084] Device Example 1
[0085] An embodiment of the present invention provides an electronic device, including:
[0086] processor; and,
[0087] A memory is arranged to store computer executable instructions, which when executed cause the processor to perform the steps of the above method embodiments.
[0088] Device Example 2
[0089] An embodiment of the present invention provides a storage medium for storing computer-executable instructions, wherein the computer-executable instructions implement the steps of the above method embodiment when executed.
[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A carbon asset assessment method based on online prediction, It is characterized in that include: Acquire carbon asset information data, and divide the carbon asset information data into carbon asset historical data and carbon asset latest data; Preprocessing the historical data of carbon assets and the latest data of carbon assets; The carbon asset benchmark assessment model is obtained through an offline training method to perform carbon asset assessment and obtain assessment data. According to the pre-processed latest carbon asset data, it is determined whether to revise the carbon asset benchmark assessment model. If revision is required, the carbon asset assessment is performed through the revised carbon asset benchmark assessment model to obtain assessment data.
2. The method according to claim 1, It is characterized in that The method of obtaining the carbon asset benchmark assessment model through the offline training method specifically includes: A carbon asset benchmark assessment model is constructed based on long short-term memory neural network and recurrent unit neural network. The preprocessed carbon asset historical data is divided into a training set and a validation set. The training set is trained through the offline training part and verified with the validation set.
3. The method according to claim 2, It is characterized in that The long short-term memory neural network specifically includes: An input gate, a forget gate and an output gate are set in the long short-term memory neural network, the forget gate is used to store the weight vector of old information, the input gate is used to obtain the weight vector of new information, and the output gate is used to output the candidate vector to be output.
4. The method according to claim 3, It is characterized in that The expression of the long short-term memory neural network specifically includes: Among them, X t represents the input at time t, h t represents the output at time t, h t-1 represents the output at time t-1, C t and C t-1 are the cell states at time t and t-1 respectively, is the unit state transition value, f t 、i t and O t They represent the forget gate, input gate and output gate respectively, σ is the sigmoid activation function, tanh is the tanh activation function, W fh , W ih , W oh , W ch Represents the interconnection weight matrix of forget gate, input gate, output gate, and unit state, W fx , W ix , W ox , W cx are the input weight matrices of the forget gate, input gate, output gate, and unit state, respectively, and b f 、b i 、b o 、b c They are the corresponding bias items for the forget gate, input gate, output gate, and unit state output respectively. Represents the Hadamard product of matrices.
5. The method according to claim 2, It is characterized in that The recurrent unit neural network includes an update gate and a reset gate, and the recurrent unit neural network is obtained by formula 2; Among them, z t represents the update gate input at time t, r t represents the reset gate input at time t, is the transition value, W zx , W rx , W zh , W rh Represent the weight matrices of the update gate and reset gate, W h The weight matrix representing the transition value, b z , b r Output the corresponding bias terms for the update gate and reset gate respectively.
6. The method according to claim 2, It is characterized in that The determining whether to revise the carbon asset benchmark assessment model according to the pre-processed latest carbon asset data specifically includes: When the change in the latest data of carbon assets after preprocessing is greater than a preset threshold compared with the assessment data obtained by the carbon asset benchmark assessment model, the hedging algorithm is used to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network, and backpropagation is used to update the carbon asset benchmark assessment model.
7. The method according to claim 6, It is characterized in that The use of the hedging algorithm to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network and back-propagate to update the carbon asset benchmark assessment model specifically includes: Forward propagation is performed on each module in the long short-term memory neural network and the recurrent unit neural network, and the expression is: in, represents the initial value of the short-term memory neural network and the recurrent unit neural network, represents the input value of the lth layer of the short-term memory neural network and the recurrent unit neural network, Represents the output value of the lth layer of the short-term memory neural network and the recurrent unit neural network, FC is the fully connected layer used for residual connection, concat represents vector cascade calculation, l∈(0,1,2…L), L is the number of module layers for each channel; The output of each module is calculated through the fully connected layer to obtain the prediction result of each module, and the prediction results are weighted and summed to form the final prediction output of the model: in, represents the prediction results and output weights of the lth layer of the long short-term memory neural network, represents the prediction result and output weight of the lth layer of the recurrent unit neural network, and yy represents the final output; When the error of the final output is greater than the preset threshold, the output weight is adjusted And back-propagate to update the overall model.
8. A carbon asset assessment system based on online prediction, It is characterized in that include: A data acquisition module is used to acquire carbon asset information data and divide the carbon asset information data into carbon asset historical data and carbon asset latest data; A data processing module, used for preprocessing the historical data of carbon assets and the latest data of carbon assets; Carbon asset assessment module, including training submodule, change identification submodule and online increment submodule; The training submodule is used to obtain a carbon asset benchmark assessment model through an offline training method to perform carbon asset assessment and obtain assessment data; The change identification submodule is used to obtain the change range between the latest data of pre-processed carbon assets and the assessment data obtained by the carbon asset benchmark assessment model; The online incremental submodule is used to determine whether to revise the carbon asset benchmark assessment model based on the preprocessed latest carbon asset data. If revision is required, the carbon asset assessment is performed using the revised carbon asset benchmark assessment model to obtain assessment data.
9. The system according to claim 8, It is characterized in that The training submodule is specifically used for: A carbon asset benchmark assessment model is constructed based on long short-term memory neural network and recurrent unit neural network. The preprocessed carbon asset historical data is divided into a training set and a validation set. The training set is trained through the offline training part and verified with the validation set.
10. The system according to claim 9, It is characterized in that An input gate, a forget gate and an output gate are arranged in the long short-term memory neural network. The forget gate is used to store the weight vector of old information, the input gate is used to obtain the weight vector of new information, and the output gate is used to output the candidate vector to be output.
11. The system according to claim 10, It is characterized in that The expression of the long short-term memory neural network specifically includes: Among them, X t represents the input at time t, h t represents the output at time t, h t-1 represents the output at time t-1, C t and C t-1 are the cell states at time t and t-1 respectively, is the unit state transition value, f t 、i t and O t They represent the forget gate, input gate and output gate respectively, σ is the sigmoid activation function, tanh is the tanh activation function, W fh , W ih , W oh , W ch Represents the interconnection weight matrix of forget gate, input gate, output gate, and unit state, W fx , W ix , W ox , W cx are the input weight matrices of the forget gate, input gate, output gate, and unit state, respectively, and b f 、b i 、b o 、b c They are the corresponding bias items for the forget gate, input gate, output gate, and unit state output respectively. Represents the Hadamard product of matrices.
12. The system according to claim 9, It is characterized in that The recurrent unit neural network includes an update gate and a reset gate, and the recurrent unit neural network is obtained by formula 2; Among them, z t represents the update gate input at time t, r t represents the reset gate input at time t, is the transition value, W zx , W rx , W zh , W rh Represent the weight matrices of the update gate and reset gate, W h The weight matrix representing the transition value, b z 、b r Output the corresponding bias terms for the update gate and reset gate respectively.
13. The system according to claim 9, It is characterized in that The online increment submodule is specifically used for: When the change in the latest data of carbon assets after preprocessing is greater than a preset threshold compared with the assessment data obtained by the carbon asset benchmark assessment model, the hedging algorithm is used to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network, and backpropagation is used to update the carbon asset benchmark assessment model.
14. The system according to claim 13, It is characterized in that The online incremental submodule uses a hedging algorithm to adjust the output weights of the long short-term memory neural network and the recurrent unit neural network and back-propagates to update the carbon asset benchmark assessment model, specifically including: Forward propagation is performed on each module in the long short-term memory neural network and the recurrent unit neural network, and the expression is: in, represents the initial value of the short-term memory neural network and the recurrent unit neural network, represents the input value of the lth layer of the short-term memory neural network and the recurrent unit neural network, Represents the output value of the lth layer of the short-term memory neural network and the recurrent unit neural network, FC is the fully connected layer used for residual connection, concat represents vector cascade calculation, l∈(0,1,2…L), L is the number of module layers for each channel; The output of each module is calculated through the fully connected layer to obtain the prediction result of each module, and the prediction results are weighted and summed to form the final prediction output of the model: in, represents the prediction results and output weights of the lth layer of the long short-term memory neural network, represents the prediction result and output weight of the lth layer of the recurrent unit neural network, and yy represents the final output; When the error of the final output is greater than the preset threshold, the output weight is adjusted And back-propagate to update the overall model.
15. An electronic device, include: processor; as well as, A memory arranged to store computer executable instructions, which, when executed, cause the processor to perform the steps of the carbon asset assessment method based on online prediction as described in any one of claims 1 to 7.
16. A storage medium for storing computer executable instructions, wherein the computer executable instructions, when executed, implement the steps of the carbon asset assessment method based on online prediction as described in any one of claims 1 to 7.
Citation Information
Cited By
Thermal power plant carbon dioxide control system and method based on double-neural-network adaptive feedback
CN121069775A
