CN117035464A - Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network - Google Patents
Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network Download PDFInfo
- Publication number
- CN117035464A CN117035464A CN202311085295.2A CN202311085295A CN117035464A CN 117035464 A CN117035464 A CN 117035464A CN 202311085295 A CN202311085295 A CN 202311085295A CN 117035464 A CN117035464 A CN 117035464A
- Authority
- CN
- China
- Prior art keywords
- carbon emission
- prediction model
- emission prediction
- hidden
- time sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 title claims abstract description 250
- 229910052799 carbon Inorganic materials 0.000 title claims abstract description 250
- 230000005611 electricity Effects 0.000 title claims abstract description 171
- 238000000034 method Methods 0.000 title claims abstract description 45
- 239000013598 vector Substances 0.000 claims description 125
- 238000012549 training Methods 0.000 claims description 51
- 238000004364 calculation method Methods 0.000 claims description 32
- 238000012360 testing method Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 23
- 238000005457 optimization Methods 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000010606 normalization Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 10
- 238000012163 sequencing technique Methods 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 238000004880 explosion Methods 0.000 abstract description 7
- 230000008034 disappearance Effects 0.000 abstract description 6
- 230000008569 process Effects 0.000 description 11
- 238000013528 artificial neural network Methods 0.000 description 8
- 125000004122 cyclic group Chemical group 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000002759 z-score normalization Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 229910002092 carbon dioxide Inorganic materials 0.000 description 1
- 239000001569 carbon dioxide Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
- G06Q10/06375—Prediction of business process outcome or impact based on a proposed change
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
- G06F18/2113—Selection of the most significant subset of features by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Probability & Statistics with Applications (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
According to the enterprise electricity consumption carbon emission prediction method based on the time sequence network improved circulation network, when electricity consumption data are subjected to carbon emission prediction, the electricity consumption data can be normalized according to time sequence to obtain an electricity consumption time sequence, and then a target carbon emission prediction model is determined. In the target carbon emission prediction model, the gating circulation unit network has stronger memory capacity and generalization capacity, so that the problems of gradient disappearance and gradient explosion of the model in the time sequence treatment can be avoided, and the accuracy and reliability of carbon emission prediction are improved.
Description
Technical Field
The application relates to the technical field of energy prediction, in particular to an enterprise electricity carbon emission prediction method based on a time sequence network improved circulation network.
Background
At present, the prediction methods of the electricity consumption and carbon emission of enterprises are mainly divided into two types: in the modeling based on physics, modeling and analysis are required to be carried out on physical characteristics of buildings, equipment and electric equipment, modeling complexity is high, and prediction accuracy and reliability of the model on carbon emission are low when facing complex industrial environments or sudden weather changes, while in the modeling based on data, gradient vanishing or gradient explosion is caused when the model predicts due to certain noise and uncertainty of historical electricity data in enterprises, and further accuracy and reliability of carbon emission prediction are affected.
Disclosure of Invention
The application aims to at least solve one of the technical defects, in particular to the technical defects that the gradient of an enterprise electricity carbon emission prediction model disappears or gradient explodes during prediction in the prior art, and the accuracy and the reliability of carbon emission prediction are affected.
The application provides an enterprise electricity carbon emission prediction method based on a time sequence network improved circulation network, which comprises the following steps:
Acquiring electricity consumption data of an enterprise to be predicted, and carrying out normalization processing on the electricity consumption data according to time sequencing to obtain an electricity consumption time sequence;
determining a target carbon emission prediction model, wherein the carbon emission prediction model is obtained by training a preset initial carbon emission prediction model by taking a sample electricity time sequence as a training sample and taking a carbon emission true value corresponding to the sample electricity time sequence as a sample label, and performing parameter optimization on the initial carbon emission prediction model after training by using a gating circulation unit network;
and inputting the electricity utilization time sequence into the target carbon emission prediction model to obtain the carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
Optionally, the determining the target carbon emission prediction model includes:
acquiring historical electricity consumption data of an enterprise to be predicted and a carbon emission real value corresponding to the historical electricity consumption data, and carrying out normalization processing on the historical electricity consumption data according to time sequencing to obtain a historical electricity consumption time sequence;
dividing the historical electricity utilization time sequence into a training set and a testing set, and inputting the historical electricity utilization time sequence in the training set into a preset initial carbon emission prediction model to obtain a carbon emission predicted value output by the initial carbon emission prediction model;
Iteratively training the initial carbon emission prediction model by taking a carbon emission actual value of the carbon emission prediction value result approaching to the historical electricity consumption data as a target until the initial carbon emission prediction model meets a preset training condition to obtain an intermediate carbon emission prediction model;
and inputting the historical electricity utilization time sequence in the test set into the intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain a target carbon emission prediction model.
Optionally, the initial carbon emission prediction model includes a first input layer, a first hidden layer, and a first output layer; the first hidden layer is composed of a plurality of time steps;
the step of inputting the historical electricity utilization time sequence in the training set into a preset initial carbon emission prediction model to obtain a carbon emission predicted value output by the initial carbon emission prediction model, comprises the following steps:
inputting the historical electricity utilization time sequence in the training set into the first input layer, and converting the historical electricity utilization time sequence into input vectors of all time steps by utilizing the first input layer;
performing time sequence iteration operation on each input vector through each time step in the first hidden layer to obtain a hidden vector output by the last time step;
And inputting the hidden vector to the first output layer for calculation so that the first output layer outputs a carbon emission predicted value output by the initial carbon emission prediction model.
Optionally, the calculation formula of the first hidden layer is:
h t ＝tanh(W hx x t +W hh h t-1 +b h )
wherein h is t Representing the hidden vector when the time step is t, tanh is the activation function of the first hidden layer h, W hx A weight matrix representing the first input layer x to the first hidden layer h, W hh A weight matrix representing the first hidden layer h to the first hidden layer h, x t Representing the input vector at time step t, h t-1 Representing the hidden vector, b, at time step t-1 h Representing the bias vector of the first hidden layer h.
Optionally, the calculation formula of the first output layer is:
y t ＝softmax(W oh h t +b o )
wherein y is t Representing the predicted value of carbon emission at time step t, softmax representing the activation function of the first output layer o, W oh A weight matrix representing the first hidden layer h to the output layer o, h t Representing a hidden vector, b, output by the first hidden layer h at time step t o Representing the offset vector of the first output layer o.
Optionally, the intermediate carbon emission prediction model includes a second input layer, a second hidden layer and a second output layer, the second hidden layer is composed of a plurality of time steps, and the gating cycle unit network is arranged in the second hidden layer and is composed of a reset gate and an update gate;
The step of inputting the historical electricity utilization time sequence in the test set into the intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain a target carbon emission prediction model, comprises the following steps:
inputting the historical electricity utilization time sequence in the test set into the second input layer, and converting the historical electricity utilization time sequence in the test set into input vectors of all time steps by utilizing the second input layer;
for each time step in the second hidden layer:
inputting the input vector of the time step and the hidden vector of the last time step into the reset gate and the update gate to obtain a reset state output by the reset gate and an update state output by the update gate; wherein, the hidden vector of the last time step corresponding to the first time step in the second hidden layer is 0;
determining a candidate hidden state of the time step based on the reset state and the input vector of the time step, calculating to obtain a reset vector of the time step according to the candidate hidden state and the update state, and taking the reset vector as a hidden vector of the time step;
And inputting the hidden vector of the last time step output by the second hidden layer to the second output layer for calculation until all the historical electricity utilization time sequences in the test set are calculated, and obtaining a target carbon emission prediction model.
Optionally, the calculation formula of the reset vector is:
h t ＝(1-Z t )⊙h t-1 +Z t ⊙h t ′
wherein h is t Representing the reset vector, Z, at time step t t Representing the update state of the output of the update gate when the time step is t, h t-1 Represents the hidden vector when the time step is t-1, h t ' indicates a hidden state candidate when the time step is t, and ". As shown in FIGS, the element multiplication is performed.
The application also provides an enterprise electricity carbon emission prediction device, which comprises:
the data acquisition module is used for acquiring the power utilization data of the enterprise to be predicted, and carrying out normalization processing on the power utilization data according to time sequencing to obtain a power utilization time sequence;
the model determining module is used for determining a target carbon emission prediction model, wherein the carbon emission prediction model is obtained by taking a sample electricity time sequence as a training sample, taking a carbon emission true value corresponding to the sample electricity time sequence as a sample label to train a preset initial carbon emission prediction model, and carrying out parameter optimization on the trained initial carbon emission prediction model by utilizing a gate control circulation unit network;
And the model prediction module is used for inputting the electricity utilization time sequence into the target carbon emission prediction model to obtain the carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
The present application also provides a storage medium having stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the method for predicting electrical carbon dioxide emissions for an enterprise as in any one of the embodiments above.
The present application also provides a computer device comprising: one or more processors, and memory;
the memory has stored therein computer readable instructions that, when executed by the one or more processors, perform the steps of the enterprise electricity carbon emission prediction method of any of the above embodiments.
From the above technical solutions, the embodiment of the present application has the following advantages:
according to the enterprise electricity consumption carbon emission prediction method based on the time sequence network improved circulation network, when electricity consumption data of an enterprise to be predicted is subjected to carbon emission prediction, the electricity consumption data can be normalized according to time sequence to obtain an electricity consumption time sequence, the electricity consumption data can be described in time, then a target carbon emission prediction model can be determined, and because the model takes a sample electricity consumption time sequence as a training sample, a carbon emission true value corresponding to the sample electricity consumption time sequence as a sample label, a preset initial carbon emission prediction model is trained, and parameters of the trained initial carbon emission prediction model are optimized by utilizing a gating circulation unit network, and therefore, after the electricity consumption time sequence is input into the target carbon emission prediction model, the target carbon emission prediction model can directly output a carbon emission prediction value corresponding to the electricity consumption time sequence. In the target carbon emission prediction model, the gating circulation unit network has better memory capacity and stronger generalization capacity, so that the problems of gradient disappearance and gradient explosion of the model in time sequence processing can be avoided, and the accuracy and reliability of carbon emission prediction are improved.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the application, and that other drawings can be obtained from these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of an enterprise electricity consumption carbon emission prediction method according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of a target carbon emission prediction model training method according to an embodiment of the present application;
FIG. 3 is a schematic structural diagram of an apparatus for predicting carbon emission of electricity consumption of enterprises according to an embodiment of the present application;
fig. 4 is a schematic diagram of an internal structure of a computer device according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Based on the above, the application provides the following technical scheme, and the specific scheme is as follows:
in one embodiment, as shown in fig. 1, fig. 1 is a schematic flow chart of an enterprise electricity carbon emission prediction method according to an embodiment of the present application; the embodiment of the application provides an enterprise electricity carbon emission prediction method, which specifically comprises the following steps:
s110: and acquiring electricity consumption data of the enterprise to be predicted, and carrying out normalization processing on the electricity consumption data according to time sequencing to obtain an electricity consumption time sequence.
In this step, when predicting the carbon emission, an enterprise needing to perform the carbon emission prediction may be first used as an enterprise to be predicted, then electricity consumption data of the enterprise to be predicted, which needs to perform the carbon emission prediction, may be obtained, and the electricity consumption data may be normalized according to a time sequence, so as to obtain an electricity consumption time sequence, so as to perform the carbon emission prediction based on the electricity consumption time sequence.
It can be understood that the electricity consumption data can be recorded and stored in real time by the electricity meter or the sensor, and the electricity consumption data refer to the electricity consumption, the electricity consumption time and other data of the electricity consumption of the enterprise to be predicted in any time period, and the data can be the total amount of electricity consumption of at least one electricity consumption type in industrial electricity consumption, office electricity consumption or other electricity consumption, and the electricity consumption type is not limited herein.
Specifically, when converting the electricity data into the electricity time sequence, the electricity data in the time period to be predicted can be sequenced according to the time sequence, then normalization processing can be performed on the sequenced electricity data, each data in the time sequence is mapped to a section between (0, 1) or (-1, 1), or a certain norm of each data is mapped to 1, so that dimensional differences among the data are eliminated, and the influence on the accuracy of subsequent data processing is avoided. It should be noted that, if the electricity consumption data recorded by the enterprise to be predicted is directly used as the training data of the neural network model, problems may be brought to the training process of the model, such as slow model training speed, slow convergence caused by excessive iteration times, and the like. Therefore, the model training efficiency can be improved in a normalization processing mode
In the normalization process of the power consumption data, maximum and minimum normalization or z-score normalization may be used, which is not limited herein. The maximum and minimum value normalization refers to dividing the power consumption data by the difference between the maximum value and the minimum value after subtracting the minimum value, so as to obtain normalized data; and z-score normalization refers to dividing the power consumption data by the standard deviation after subtracting the average value.
S120: and determining a target carbon emission prediction model.
In this step, after the electricity consumption time sequence is obtained in step S110, a target carbon emission prediction model may be determined, and the carbon emission prediction may be performed on the electricity consumption time sequence by using the target carbon emission prediction model, so as to obtain a corresponding carbon emission predicted value.
Specifically, the target carbon emission prediction model of the present application refers to a model that performs carbon emission prediction on an input electricity consumption time sequence and outputs a carbon emission predicted value after obtaining the carbon emission predicted value, where the target carbon emission prediction model may use different types of sample electricity consumption time sequences as training samples when performing model training, obtain a carbon emission actual value counted in a corresponding time period of each training sample from a storage module of an enterprise to be predicted, and use the obtained carbon emission actual value as a sample tag. And then, inputting the training sample into a preset initial carbon emission prediction model for forward propagation to train the model, and carrying out reverse propagation on the model based on a sample label to carry out parameter tuning on the model, wherein when the model meets certain training conditions or convergence conditions of parameters, if the iteration number reaches a set value, the model is regarded as training completion. And then, introducing a gating circulation unit network into the model, further optimizing parameters of the initial carbon emission prediction model after training by using the gating circulation unit network, and finally, taking the optimized model as a final target carbon emission prediction model.
It can be understood that the gating circulation unit network is a circulation neural network structure of a gating mechanism, which can effectively control the influence of the hidden state at the previous moment on the input at the current moment, avoid the problems of gradient disappearance and gradient explosion, and can selectively update and reserve the hidden state at the previous moment, thereby realizing better memory capacity.
In addition, the application can store the target carbon emission prediction model after optimization, so that the carbon emission prediction model stored in advance can be directly called to predict the carbon emission of the electricity utilization time sequence in the follow-up carbon emission prediction. Further, when the initial carbon emission prediction model is trained, a recurrent neural network (Rerrent Neural Network, RNN) can be selected as a preset initial carbon emission prediction model for improvement and training, and the RNN is a neural network structure with time sequence processing capability, and unlike a traditional feedforward neural network, the RNN can process time sequence data with any length and transmit history information to a network state at the current moment, so that the processing and memorizing of the time sequence information are realized. Other neural networks of similar structures can be used as the initial carbon emission prediction model preset in the present application, and are not limited herein.
S130: and inputting the electricity utilization time sequence into a target carbon emission prediction model to obtain a carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
In the step, after the target carbon emission prediction model is determined in the step S120, the electricity consumption time sequence is input into the target carbon emission prediction model, and the carbon emission prediction is performed on the electricity consumption time sequence through the target carbon emission prediction model.
Specifically, when the electricity time sequence is predicted, iterative operation can be performed on each data in the electricity time sequence, and because the target carbon emission prediction model has memory capacity in the process of performing iterative operation on each data in the electricity time sequence, the series connection and the memory of the information of the electricity time sequence can be realized, so that a carbon emission predicted value corresponding to the electricity time sequence can be obtained. In the iterative operation process, the hidden vector of each operation process can be optimized and adjusted by using the gating loop unit network, so that the problem of long-time dependence in the electricity utilization time sequence prediction process is solved, and the accuracy and reliability of the carbon emission predicted value can be improved.
In the above embodiment, when the carbon emission prediction is performed on the electricity consumption data of the enterprise to be predicted, the electricity consumption data may be normalized according to the time sequence to obtain the electricity consumption time sequence, so as to describe the electricity consumption data in time, and then the target carbon emission prediction model may be determined. In the target carbon emission prediction model, the gating circulation unit network has better memory capacity and stronger generalization capacity, so that the problems of gradient disappearance and gradient explosion of the model in time sequence processing can be avoided, and the accuracy and reliability of carbon emission prediction are improved.
In one embodiment, as shown in fig. 2, fig. 2 is a schematic flow chart of a target carbon emission prediction model training method according to an embodiment of the present application; in fig. 2, the determining the target carbon emission prediction model in step S120 may include:
S121: and acquiring historical electricity consumption data of the enterprise to be predicted and a carbon emission real value corresponding to the historical electricity consumption data, and carrying out normalization processing on the historical electricity consumption data according to time sequencing to obtain a historical electricity consumption time sequence.
S122: dividing the historical electricity utilization time sequence into a training set and a testing set, and inputting the historical electricity utilization time sequence in the training set into a preset initial carbon emission prediction model to obtain a carbon emission predicted value output by the initial carbon emission prediction model.
S123: and carrying out iterative training on the initial carbon emission prediction model by taking a carbon emission actual value, of which the carbon emission prediction value result approaches to the historical electricity consumption data, as a target until the initial carbon emission prediction model meets preset training conditions, so as to obtain an intermediate carbon emission prediction model.
S124: and (3) inputting the historical electricity utilization time sequence in the test set into an intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain a target carbon emission prediction model.
In this embodiment, when determining the target carbon emission prediction model, a corresponding neural network model may be selected as a preset initial carbon emission prediction model for improvement and training. Before model training, historical electricity consumption data of an enterprise to be predicted and a carbon emission true value corresponding to the historical electricity consumption data can be obtained, after the historical electricity consumption data is converted into an electricity consumption time sequence, the electricity consumption time sequence is divided into a training set and a testing set, so that after the historical electricity consumption data in the training set is input into a preset initial carbon emission prediction model, a carbon emission predicted value output by the initial carbon emission prediction model can be obtained, and then the carbon emission prediction model can be trained by taking the carbon emission predicted value approaching to the carbon emission true value corresponding to the historical electricity consumption data in the training set as a target. And then, the historical electricity utilization time sequence in the test set can be input into the intermediate carbon emission prediction model, and the intermediate carbon emission prediction model is subjected to parameter optimization by utilizing the gating circulation unit network, so that the intermediate carbon emission prediction model after optimization is used as a target carbon emission prediction model.
In a specific implementation, the present application may use MSE (Mean Squared Error, mean square error) to characterize the difference between the predicted value of carbon emissions and the actual value of carbon emissions, where a larger MSE indicates a larger deviation between the predicted value of carbon emissions and the actual value of carbon emissions, and where a smaller MSE indicates a closer predicted value of carbon emissions to the actual value of carbon emissions, and where the model has a higher prediction accuracy. Wherein, the calculation formula of MSE is as follows:
wherein t represents the number of samples in the training set, y t Representing the actual value of carbon emissions for the ith sample,representing the predicted value of carbon emission for the i-th sample.
Furthermore, the application can also preprocess the historical electricity data before normalizing the historical electricity data into the historical electricity time sequence, such as data cleaning, data smoothing and the like, wherein the data cleaning can remove abnormal values and noise in the historical electricity data, and the data smoothing can remove random fluctuation values in the historical electricity data so as to make the historical electricity data more stable and smooth, thereby improving the efficiency and performance of model training.
In one embodiment, the initial carbon emission prediction model in step S122 includes a first input layer, a first hidden layer, and a first output layer; the first hidden layer is composed of a plurality of time steps; the method for obtaining the carbon emission prediction value output by the initial carbon emission prediction model comprises the following steps of:
S1221: the historical electricity utilization time sequence in the training set is input into a first input layer, and the historical electricity utilization time sequence is converted into input vectors of all time steps by utilizing the first input layer.
S1222: and performing time sequence iterative operation on each input vector through each time step in the first hidden layer to obtain the hidden vector output by the last time step.
S1223: and inputting the hidden vector into a first output layer for calculation so that the first output layer outputs a carbon emission predicted value output by the initial carbon emission prediction model.
In this embodiment, when the historical electricity time sequence in the training set is input into the preset initial carbon emission prediction model, the historical electricity time sequence may be input into the first input layer of the initial carbon emission prediction model, at this time, the input layer may convert the historical electricity time sequence into input vectors of each time step, and output each input vector into the first hidden layer, and then the first hidden layer may sequentially perform time sequence iterative operation on each input vector by using each time step, so as to obtain a hidden vector output by the last time step, and input the hidden vector into the first output layer for calculation, so that the first output layer outputs a carbon emission predicted value output by the initial carbon emission prediction model.
It can be understood that the initial carbon emission prediction model is used as a recurrent neural network model, has a memory function, and is in a fully connected state between the first input layer and the first hidden layer and between the first hidden layer and the first output layer, and the hidden layer is a self-circulation structure, so that all relevant data which are calculated at present can be captured by the initial carbon emission prediction model. In addition, the first hidden layer executes the same task on each input vector in the historical electricity utilization time sequence, and the result output after execution is influenced by the front operation, so that the front information can be transferred to the calculation of the rear time step, and the memorizing effect is achieved.
Specifically, the first input layer is used as an input end of the initial carbon emission prediction model and has the effects of processing and memorizing a time sequence, and each data in the historical electricity utilization time sequence can be mapped into a vector with a fixed length, so that an input vector of each time step is obtained, and basic data is provided for the prediction of a subsequent model; then the first hidden layer can utilize each time step to sequentially carry out time sequence iterative operation on the input vector output by the first input layer, and generate a hidden vector corresponding to each time step, as the operation time steps increase, the historical data and the input data contained in the hidden vector are more and more, until the last time step, the hidden vector containing all the historical data and the input data can be obtained, and at the moment, the hidden vector can be used as the output of the first hidden layer and transmitted to the first output layer for processing and prediction; the first output layer may perform a weighting operation based on the hidden vector after receiving the hidden vector, to convert the hidden vector into a carbon emission prediction value, and use the carbon emission prediction value as an output value of the initial carbon emission prediction model.
In one embodiment, the calculation formula of the first hidden layer in step S1222 is:
h t ＝tanh(W hx x t +W hh h t-1 +b h )
wherein h is t Representing the hidden vector when the time step is t, tanh is the activation function of the first hidden layer h, W hx A weight matrix representing the first input layer x to the first hidden layer h, W hh Represent the firstWeight matrix from hidden layer h to first hidden layer h, x t Representing the input vector at time step t, h t-1 Representing the hidden vector, b, at time step t-1 h Representing the bias vector of the first hidden layer h.
In this embodiment, when determining the calculation formula of the first hidden layer of the present application, the first input layer may be denoted by x, the first hidden layer may be denoted by h, and then the weight matrix W from the first input layer x to the first hidden layer h may be determined hx And a weight matrix W from the first hidden layer h to the first hidden layer h hh At the same time, the bias vector b of the preset first hidden layer can also be set for the performance of the initial carbon emission prediction model h And further determining a calculation formula of the first hidden layer according to the association between the data, and taking the calculation formula as an execution task of each time step in the first hidden layer.
It will be appreciated that in determining the bias vector b for the first hidden layer h The method can be determined by adopting a random initialization or optimization algorithm and the like. When the bias vector is determined by adopting random initialization, the initial state of the first hidden layer can be made to be random, so that the training and convergence effects of the model can be improved; when the bias vector is determined by an optimization algorithm, algorithms such as gradient descent, random gradient descent, adam and the like can be adopted, and the algorithms can adjust the bias vector according to the change of the loss function.
Specifically, after each time step in the first hidden layer receives the input vector of the current time step and the hidden vector of the previous time step, a calculation formula can be used for calculating to obtain the hidden vector of the current time step, and the hidden vector of the current time step is transferred to the next time step as the input of the calculation formula, so that along with the iterative operation of each time step in the first hidden layer, the last time step can obtain the hidden vector containing all data information. Further, when the first hidden layer performs the first time step calculation, the input hidden vector of the last time step may be a random initial value or may be 0, which is specifically determined according to the parameters and performance of the initial carbon emission prediction model, and is not limited herein.
In one embodiment, the calculation formula of the first output layer in step S1223 is:
y t ＝soffmax(W oh h t +b o )
wherein y is t Representing the predicted value of carbon emission at time step t, soffmax represents the activation function of the first output layer o, W oh A weight matrix representing the first hidden layer h to the output layer o, h t Representing a hidden vector, b, output by the first hidden layer h at time step t o Representing the offset vector of the first output layer o.
In this embodiment, when determining the calculation formula of the first output layer of the present application, the first output layer may be represented by o, the first hidden layer may be represented by h, and then the weight matrix W from the first hidden layer h to the first output layer o may be determined oh At the same time, the bias vector b of the preset first output layer can also be set for the performance of the initial carbon emission prediction model o Further, a calculation formula of the first output layer may be determined based on the association of the respective data.
In one embodiment, the intermediate carbon emission prediction model in step S123 includes a second input layer, a second hidden layer, and a second output layer, where the second hidden layer is composed of a plurality of time steps, and the gating cycle unit network is disposed in the second hidden layer and is composed of a reset gate and an update gate; the method for obtaining the target carbon emission prediction model includes the steps of inputting a historical electricity utilization time sequence in a test set into an intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain the target carbon emission prediction model, wherein the method comprises the following steps:
s1231: and inputting the historical electricity utilization time sequence in the test set into a second input layer, and converting the historical electricity utilization time sequence in the test set into input vectors of all time steps by utilizing the second input layer.
S1232: the input vector of the time step and the hidden vector of the last time step are input into a reset gate and an update gate to obtain a reset state output by the reset gate and an update state output by the update gate; wherein, the hidden vector of the last time step corresponding to the first time step in the second hidden layer is 0.
S1233: and determining a candidate hidden state of the time step based on the reset state and the input vector of the time step, calculating to obtain a reset vector of the time step according to the candidate hidden state and the updated state, and taking the reset vector as the hidden vector of the time step.
S1234: and inputting the hidden vector of the last time step output by the second hidden layer to the second output layer for calculation until all the historical electricity utilization time sequences in the test set are calculated, and obtaining the target carbon emission prediction model.
In this embodiment, after the intermediate carbon emission prediction model is obtained, a gating cyclic unit network may be introduced into a second hidden layer of the intermediate carbon emission prediction model, and a historical electricity time sequence in a test set is input into a second input layer, so that the historical electricity time sequence in the test set is converted into input vectors of each time step through the second input layer, then a reset gate in the gating cyclic unit network and updating and input vectors of each time step may be used to optimize hidden vectors of each time step in the second hidden layer until a hidden vector of the last time step is obtained, and the hidden vector is used as an input of a second output layer, so that the second output layer calculates a carbon emission prediction value corresponding to the historical electricity time sequence in the test set, and when all the historical electricity time sequences in the test set are calculated, optimization of the intermediate carbon emission test model is completed, and the optimized intermediate carbon emission test model is used as the target carbon emission prediction model.
It should be noted that the gating cyclic unit network in the application is an improved cyclic neural network structure, introduces a reset gate and an update gate mechanism, solves the problems of gradient disappearance and gradient explosion when the traditional cyclic neural network processes long sequence data, and has better memory capacity and stronger generalization capacity.
Specifically, when the network circulation unit network is utilized to optimize the intermediate carbon emission prediction model, for each time step in the second hidden layer, the input vector of the time step and the hidden vector of the last time step can be input into the reset gate and the update gate to obtain the reset state of the output of the reset gate and the update state of the output of the update gate; the hidden vector of the last time step corresponding to the first time step in the second hidden layer is 0, the candidate hidden state of the time step is determined based on the reset state and the input vector of the time step, the reset vector of the time step is obtained through calculation according to the candidate hidden state and the update state, and the reset vector is used as the hidden vector of the time step. The calculation result of the network circulation unit network includes a reset state of the output of the reset gate, an update state of the output of the update gate, a candidate hidden state and a reset vector, wherein the reset state and the update state can be calculated by a sigmoid function, and the candidate hidden state and the reset vector can be calculated by a tanh function.
In one specific implementation, the formulas for resetting the reset state of the gate and updating the update state of the gate are as follows:
wherein R is t Indicating the reset state of the reset gate, Z t Representing the update state of the update gate, delta represents the sigmoid function,weight matrix representing reset gates, +.>Weight matrix, x, representing update gates t Representing the input vector at time step t, h t-1 Representing the hidden vector at time step t-1.
In addition, the calculation formula of the candidate hidden state when the time step is t is specifically as follows:
h′ t-1 ＝h t-1 ⊙R t
in the formula, h' t-1 Indicating candidate hidden states when the time step is t-1, h t-1 Indicating the hidden vector when the time step is t-1, as indicated by the element multiplication, h t ' indicates a candidate hidden state when the time step is t.
It will be appreciated that the gating loop network incorporates both reset and update gating mechanisms, which can be used to control the effect of the hidden vector of the previous time step on the input vector of the current time step. The reset gate is used for controlling the influence degree of the hidden vector of the previous time step on the input vector of the current time step, and the update gate is used for controlling the influence degree of the hidden vector of the previous time step and the input vector of the current time step on the hidden vector of the current time step. The mechanism can effectively control updating and transferring of the hidden vector, thereby realizing effective processing and memorizing of time sequence information.
In one embodiment, the calculation formula of the reset vector in step S1233 is:
h t ＝(1-Z t )⊙h t-1 +Z t ⊙h t ′
wherein h is t Representing the reset vector, Z, at time step t t Representing the update state of the output of the update gate when the time step is t, h t-1 Represents the hidden vector when the time step is t-1, h t ' indicates a hidden state candidate when the time step is t, and ". As shown in FIGS, the element multiplication is performed.
In this embodiment, after the reset state of the reset gate and the candidate hidden state when the time step is t are obtained by calculation, the calculation formula of the reset vector when the time step is t may be determined according to the reset state of the reset gate, the candidate hidden state when the time step is t and the hidden vector when the time step is t-1.
It will be appreciated that in the above calculation formula of the reset vector, (1-Z t )⊙h t-1 Indicating selective forgetting of hidden vectors of the last time step, Z t ⊙h t ' means that the candidate hidden state of the current time step is selectively memorized. Wherein Z is t The value range of (2) is [0,1 ]]When Z is t Approaching 1, a long-term dependence is indicated, when Z t Approaching 0, this indicates that some unimportant historical information in the hidden vector of the last time step is forgotten.
The following describes the device for predicting the carbon emission of electricity consumption of an enterprise, and the device for predicting the carbon emission of electricity consumption of an enterprise and the method for predicting the carbon emission of electricity consumption of an enterprise described in the following may be referred to correspondingly.
In one embodiment, as shown in fig. 3, fig. 3 is a schematic structural diagram of an enterprise electricity carbon emission prediction device according to an embodiment of the present application; the application also provides an enterprise electricity carbon emission prediction device, which comprises a data acquisition module 210, a model determination module 220 and a model prediction module 230, and specifically comprises the following steps:
the data obtaining module 210 is configured to obtain power utilization data of an enterprise to be predicted, and perform normalization processing on the power utilization data according to time sequencing, so as to obtain a power utilization time sequence.
The model determining module 220 is configured to determine a target carbon emission prediction model, where the carbon emission prediction model is obtained by training a preset initial carbon emission prediction model with a sample electricity time sequence as a training sample and with a carbon emission true value corresponding to the sample electricity time sequence as a sample tag, and performing parameter optimization on the trained initial carbon emission prediction model by using a gating circulation unit network.
The model prediction module 230 is configured to input the electricity consumption time sequence to a target carbon emission prediction model, and obtain a carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
In the above embodiment, when the carbon emission prediction is performed on the electricity consumption data of the enterprise to be predicted, the electricity consumption data may be normalized according to the time sequence to obtain the electricity consumption time sequence, so as to describe the electricity consumption data in time, and then the target carbon emission prediction model may be determined. In the target carbon emission prediction model, the gating circulation unit network has better memory capacity and stronger generalization capacity, so that the problems of gradient disappearance and gradient explosion of the model in time sequence processing can be avoided, and the accuracy and reliability of carbon emission prediction are improved.
In one embodiment, the present application also provides a storage medium having stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the method for predicting electrical carbon emissions for an enterprise as in any one of the above embodiments.
In one embodiment, the present application also provides a computer device having stored therein computer readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the method for predicting electrical carbon emissions for an enterprise as in any one of the embodiments above.
Schematically, as shown in fig. 4, fig. 4 is a schematic internal structure of a computer device according to an embodiment of the present application, and the computer device 300 may be provided as a server. Referring to FIG. 4, computer device 300 includes a processing component 302 that further includes one or more processors, and memory resources represented by memory 301, for storing instructions, such as applications, executable by processing component 302. The application program stored in the memory 301 may include one or more modules each corresponding to a set of instructions. Further, the processing component 302 is configured to execute instructions to perform the enterprise electricity carbon emission prediction method of any of the embodiments described above.
The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input output (I/O) interface 305. The computer device 300 may operate based on an operating system stored in memory 301, such as Windows Server TM, mac OS XTM, unix TM, linux TM, free BSDTM, or the like.
It will be appreciated by persons skilled in the art that the architecture shown in fig. 4 is merely a block diagram of some of the architecture relevant to the present inventive arrangements and is not limiting as to the computer device to which the present inventive arrangements are applicable, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
Finally, it is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the present specification, each embodiment is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and may be combined according to needs, and the same similar parts may be referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. An enterprise electricity carbon emission prediction method based on a time sequence network improved circulation network, which is characterized by comprising the following steps:
acquiring electricity consumption data of an enterprise to be predicted, and carrying out normalization processing on the electricity consumption data according to time sequencing to obtain an electricity consumption time sequence;
determining a target carbon emission prediction model, wherein the carbon emission prediction model is obtained by training a preset initial carbon emission prediction model by taking a sample electricity time sequence as a training sample and taking a carbon emission true value corresponding to the sample electricity time sequence as a sample label, and performing parameter optimization on the initial carbon emission prediction model after training by using a gating circulation unit network;
And inputting the electricity utilization time sequence into the target carbon emission prediction model to obtain the carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
2. The method for predicting electrical carbon emissions of an enterprise as claimed in claim 1, wherein the determining the target carbon emission prediction model comprises:
acquiring historical electricity consumption data of an enterprise to be predicted and a carbon emission real value corresponding to the historical electricity consumption data, and carrying out normalization processing on the historical electricity consumption data according to time sequencing to obtain a historical electricity consumption time sequence;
dividing the historical electricity utilization time sequence into a training set and a testing set, and inputting the historical electricity utilization time sequence in the training set into a preset initial carbon emission prediction model to obtain a carbon emission predicted value output by the initial carbon emission prediction model;
iteratively training the initial carbon emission prediction model by taking a carbon emission actual value of the carbon emission prediction value result approaching to the historical electricity consumption data as a target until the initial carbon emission prediction model meets a preset training condition to obtain an intermediate carbon emission prediction model;
and inputting the historical electricity utilization time sequence in the test set into the intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain a target carbon emission prediction model.
3. The method of claim 2, wherein the initial carbon emission prediction model comprises a first input layer, a first hidden layer, and a first output layer; the first hidden layer is composed of a plurality of time steps;
the step of inputting the historical electricity utilization time sequence in the training set into a preset initial carbon emission prediction model to obtain a carbon emission predicted value output by the initial carbon emission prediction model, comprises the following steps:
inputting the historical electricity utilization time sequence in the training set into the first input layer, and converting the historical electricity utilization time sequence into input vectors of all time steps by utilizing the first input layer;
performing time sequence iteration operation on each input vector through each time step in the first hidden layer to obtain a hidden vector output by the last time step;
and inputting the hidden vector to the first output layer for calculation so that the first output layer outputs a carbon emission predicted value output by the initial carbon emission prediction model.
4. The method for predicting carbon emissions for electrical use in an enterprise of claim 3, wherein the first hidden layer has a calculation formula:
h t ＝tanh(W hx x t +W hh h t-1 +b h )
Wherein h is t Representing the hidden vector when the time step is t, tanh is the activation function of the first hidden layer h, W hx A weight matrix representing the first input layer x to the first hidden layer h, W hh A weight matrix representing the first hidden layer h to the first hidden layer h, x t Representing the input vector at time step t, h t-1 Representing the hidden vector, b, at time step t-1 h Representing the bias vector of the first hidden layer h.
5. The method for predicting carbon emissions for electrical use in an enterprise of claim 3, wherein the first output layer has a calculation formula:
y t ＝softmax(W oh h t +b o )
wherein y is t Representing the predicted value of carbon emission at time step t, softmax representing the activation function of the first output layer o, W oh A weight matrix representing the first hidden layer h to the output layer o, h t Representing a hidden vector, b, output by the first hidden layer h at time step t o Representing the offset vector of the first output layer o.
6. The method for predicting the carbon emission of electricity consumption of an enterprise according to claim 2, wherein the intermediate carbon emission prediction model comprises a second input layer, a second hidden layer and a second output layer, the second hidden layer is composed of a plurality of time steps, and the gating cycle unit network is arranged in the second hidden layer and is composed of a reset gate and an update gate;
The step of inputting the historical electricity utilization time sequence in the test set into the intermediate carbon emission prediction model, and carrying out parameter optimization on the intermediate carbon emission prediction model by utilizing a gating circulation unit network to obtain a target carbon emission prediction model, comprises the following steps:
inputting the historical electricity utilization time sequence in the test set into the second input layer, and converting the historical electricity utilization time sequence in the test set into input vectors of all time steps by utilizing the second input layer;
for each time step in the second hidden layer:
inputting the input vector of the time step and the hidden vector of the last time step into the reset gate and the update gate to obtain a reset state output by the reset gate and an update state output by the update gate; wherein, the hidden vector of the last time step corresponding to the first time step in the second hidden layer is 0;
determining a candidate hidden state of the time step based on the reset state and the input vector of the time step, calculating to obtain a reset vector of the time step according to the candidate hidden state and the update state, and taking the reset vector as a hidden vector of the time step;
And inputting the hidden vector of the last time step output by the second hidden layer to the second output layer for calculation until all the historical electricity utilization time sequences in the test set are calculated, and obtaining a target carbon emission prediction model.
7. The method for predicting carbon emissions for enterprises as set forth in claim 6, wherein the calculation formula of the reset vector is:
h t ＝(1-Z t )⊙h t-1 +Z t ⊙h t ′
wherein h is t Representing the reset vector, Z, at time step t t Representing the update state of the output of the update gate when the time step is t, h t-1 Represents the hidden vector when the time step is t-1, h t ' indicates a hidden state candidate when the time step is t, and ". As shown in FIGS, the element multiplication is performed.
8. An enterprise electricity carbon emission prediction device, characterized by comprising:
the data acquisition module is used for acquiring the power utilization data of the enterprise to be predicted, and carrying out normalization processing on the power utilization data according to time sequencing to obtain a power utilization time sequence;
the model determining module is used for determining a target carbon emission prediction model, wherein the carbon emission prediction model is obtained by taking a sample electricity time sequence as a training sample, taking a carbon emission true value corresponding to the sample electricity time sequence as a sample label to train a preset initial carbon emission prediction model, and carrying out parameter optimization on the trained initial carbon emission prediction model by utilizing a gate control circulation unit network;
And the model prediction module is used for inputting the electricity utilization time sequence into the target carbon emission prediction model to obtain the carbon emission predicted value of the enterprise to be predicted, which is output by the target carbon emission prediction model.
9. A storage medium, characterized by: the storage medium having stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the enterprise electricity carbon emission prediction method of any one of claims 1 to 7.
10. A computer device, comprising: one or more processors, and memory;
stored in the memory are computer readable instructions which, when executed by the one or more processors, perform the steps of the enterprise electricity carbon emission prediction method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311085295.2A CN117035464A (en) | 2023-08-25 | 2023-08-25 | Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311085295.2A CN117035464A (en) | 2023-08-25 | 2023-08-25 | Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117035464A true CN117035464A (en) | 2023-11-10 |
Family
ID=88629883
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311085295.2A Pending CN117035464A (en) | 2023-08-25 | 2023-08-25 | Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117035464A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117391258A (en) * | 2023-12-08 | 2024-01-12 | 深圳碳中和生物燃气股份有限公司 | Method, device, equipment and storage medium for predicting negative carbon emission |
-
2023
- 2023-08-25 CN CN202311085295.2A patent/CN117035464A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117391258A (en) * | 2023-12-08 | 2024-01-12 | 深圳碳中和生物燃气股份有限公司 | Method, device, equipment and storage medium for predicting negative carbon emission |
| CN117391258B (en) * | 2023-12-08 | 2024-03-15 | 深圳碳中和生物燃气股份有限公司 | Method, device, equipment and storage medium for predicting negative carbon emission |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113777496B (en) | Lithium ion battery residual life prediction method based on time convolution neural network | |
| CN110909926A (en) | TCN-LSTM-based solar photovoltaic power generation prediction method | |
| CN111832825B (en) | Wind power prediction method and system integrating long-term memory network and extreme learning machine | |
| CN113852432B (en) | Spectrum Prediction Sensing Method Based on RCS-GRU Model | |
| CN117035464A (en) | Enterprise electricity consumption carbon emission prediction method based on time sequence network improved circulation network | |
| Wang et al. | Multi-step-ahead time series prediction method with stacking LSTM neural network | |
| CN114239991A (en) | Building heat supply load prediction method, device and equipment based on data driving | |
| CN115018193A (en) | Time series wind energy data prediction method based on LSTM-GA model | |
| CN116541176A (en) | Optimization method and optimization device for computing power resource allocation, electronic equipment and medium | |
| CN117319231A (en) | Network flow prediction method and system for optimizing VMD (virtual model device) based on dung beetle algorithm | |
| CN113093014B (en) | Online collaborative estimation method and system for SOH and SOC based on impedance parameters | |
| Li et al. | State of charge estimation of lithium-ion batteries based on PSO-TCN-Attention neural network | |
| CN117312835A (en) | Fuel cell performance prediction method based on mixed probability data driving model | |
| CN116610416A (en) | Load prediction type elastic expansion system and method based on Kubernetes | |
| CN114169416B (en) | Short-term load prediction method based on migration learning under small sample set | |
| CN116125279A (en) | Method, device, equipment and storage medium for determining battery health state | |
| CN113610665B (en) | Wind power generation power prediction method based on multi-delay output echo state network | |
| CN113240181B (en) | Rolling simulation method and device for reservoir dispatching operation | |
| CN116227650A (en) | Lithium battery temperature distribution prediction model construction method and model based on orthogonal enhancement type local maintenance projection algorithm | |
| CN114895206A (en) | Lithium ion battery SOH estimation method based on RBF neural network of improved wolf optimization algorithm | |
| CN114995106A (en) | PID self-tuning method, device and equipment based on improved wavelet neural network | |
| CN114330119A (en) | Deep learning-based pumped storage unit adjusting system identification method | |
| CN113112092A (en) | Short-term probability density load prediction method, device, equipment and storage medium | |
| Bizjak et al. | Time series prediction for EMS with machine learning | |
| Li et al. | A Model-Based Exploration Policy in Deep Q-Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |