CN116882580A - Public institution-oriented LSTM circulating neural network carbon emission prediction method - Google Patents
Public institution-oriented LSTM circulating neural network carbon emission prediction method Download PDFInfo
- Publication number
- CN116882580A CN116882580A CN202310893182.9A CN202310893182A CN116882580A CN 116882580 A CN116882580 A CN 116882580A CN 202310893182 A CN202310893182 A CN 202310893182A CN 116882580 A CN116882580 A CN 116882580A
- Authority
- CN
- China
- Prior art keywords
- carbon emission
- model
- lstm
- public
- representing
- 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
- 229910052799 carbon Inorganic materials 0.000 title claims abstract description 146
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 title claims abstract description 142
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 24
- 238000004364 calculation method Methods 0.000 claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 25
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims abstract description 12
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims abstract description 12
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 238000003062 neural network model Methods 0.000 claims abstract description 5
- 238000005265 energy consumption Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 10
- 125000004122 cyclic group Chemical group 0.000 claims description 8
- 230000005611 electricity Effects 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 7
- 210000004027 cell Anatomy 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000010248 power generation Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 239000003245 coal Substances 0.000 claims description 3
- 230000007774 longterm Effects 0.000 claims description 3
- 238000011425 standardization method Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000010276 construction Methods 0.000 abstract description 4
- 238000003763 carbonization Methods 0.000 abstract 1
- 238000004590 computer program Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000006872 improvement Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of carbon emission prediction, in particular to an LSTM circulating neural network carbon emission prediction method facing public institutions. Firstly, constructing a single-feature input model of long-time scale carbon emission and a multi-feature input model based on carbon emission influence factors by utilizing an LSTM neural network model; then training the predictive model with the training set and the test set, using MAPE, RMSE, R 2 And evaluating the prediction model, correcting model parameters, and continuously optimizing to obtain an optimal prediction model to predict the fluctuation trend of the discharge amount in the next period. The invention establishes a carbon emission prediction technology aiming at public institutions, can provide carbon emission index calculation and management capability for public institutions, and provides effective and highly-credible data support for low-carbon emission of public institutions; effectively being institutional lowThe carbonization construction provides an improved thought in time, and accuracy and scientificity of public institution carbon emission prediction are improved.
Description
Technical Field
The invention relates to the technical field of carbon emission prediction, in particular to an LSTM circulating neural network carbon emission prediction method facing public institutions.
Background
In recent years, the importance of carbon emission research in public institutions has been increasingly developed under the promotion of smart city construction. Currently, carbon emission calculation methods for public institutions are roughly classified into a mass balance method, an actual observation method and an emission factor method. Among them, the emission factor method is popular, and is the most dominant means for verifying the carbon emission of users. The emission factor method is basically based on the product of the carbon emission factor of each carbon emission source and its activity level as the carbon emission amount of such emission source. This conventional carbon emission accounting method is too dependent on various energy consumption data, and requires a lot of manpower and material resources. Along with the development of machine learning technology, the combination application of the carbon emission accounting method and the machine learning technology is gradually wide, a data driving model is established through training a large number of input/output processes to describe the carbon emission generation process as a research hotspot, and typical modeling algorithms comprise a neural network, a support vector machine, a fuzzy tree and the like. However, the neural network model has the problems of over fitting, long training time and the like, the support vector machine is suitable for modeling of small samples, and the function parameters and the normalization parameters in the algorithm lack an effective and unified determination method, and the algorithms have limitations and damage the time sequence characteristics of carbon emission.
In the field of carbon emission research of public institutions, the application strength of data analysis advantages and prediction performance advantages of machine learning technology is insufficient. In view of this, we propose a LSTM cyclic neural network carbon emission prediction method for public institutions, which predicts the carbon emission of public institutions by analyzing the mapping relationship between the carbon emission and energy data, so as to provide the public institutions with carbon emission index calculation and management capabilities, and provide the public institutions with effective and highly reliable data support for the low-carbon emission.
Disclosure of Invention
The invention aims to provide an LSTM circulating neural network carbon emission prediction method facing public institutions, so as to solve the problems in the background art.
To solve the technical problems, the invention aims atFirstly, constructing a single-feature input model of long-time scale carbon emission and a multi-feature input model based on carbon emission influence factors by utilizing the advantage that an LSTM neural network model can process long-time sequences; then training the predictive model with the training set and the test set, using MAPE, RMSE, R 2 Evaluating the prediction model, correcting model parameters, and continuously optimizing to obtain an optimal prediction model to predict the fluctuation trend of the emission in the next period; the method specifically comprises the following steps:
s1, data acquisition: firstly, determining an public institution to be analyzed, acquiring basic information of the public institution, energy consumption data of the public institution in a certain period and carbon emission data of the public institution in the period, and classifying and processing the carbon emission data;
s2, analysis of total carbon emission: constructing a public institution based on the carbon emission and the generated energy of the power grid in the area of the public institution, and constructing an electric power average carbon emission coefficient calculation model based on a time sequence;
s3, carbon emission accounting: firstly, classifying according to attribute characteristics of public institutions, and accounting carbon emission of different types of public institutions; and then accounting the carbon emission of each energy source in the calculation period aiming at different energy source types, consumption and carbon emission factors thereof. Comprehensively obtaining carbon emission time sequences of public institutions of different types and carbon emission time sequences of public institutions of different energy consumption types;
s4, determining an LSTM neuron structure, and designing an LSTM model network structure; establishing a structure of an LSTM model, selecting a required loss function, and initializing model parameters; the data input needs to be initialized;
s5, training the accuracy of the prediction model. After data input, the estimated value of the model is solved through forward calculation, and the average absolute percentage error (MAPE), the Root Mean Square Error (RMSE) and the deterministic correlation coefficient (R) are selected 2 ) Calculating an estimated value and a true value to measure the effect of model prediction on unknown samples;
s6, determining an optimal prediction model: and continuously circulating the steps S4-S5, optimally adjusting parameters of the LSTM model, establishing a proper LSTM model, and predicting carbon emission of the public institution by using the optimal LSTM model.
In the step S1, in the data acquisition, firstly, determining an public institution to be analyzed, and acquiring basic information of the public institution, including building position, building area, usage type and number of buildings, so as to determine attribute characteristics of the public institution;
the energy consumption data of the public institution comprise coal consumption, oil consumption, gas consumption, electricity consumption and heat consumption data in a certain period so as to determine the energy consumption characteristics of the public institution, and the carbon emission and the electricity generation of the power grid of the area where the public institution is located;
the carbon emission data is classified, namely, the carbon emission data of the public institutions in the month, the quarter and the year are classified from the two viewpoints of the application attribute of the public institutions and the application energy characteristics of the public institutions.
As a further improvement of the present technical solution, in the step S2, the expression of the electric power average carbon emission coefficient calculation model based on the time series is:
wherein R is t Representing carbon emission generated per unit power generation amount in a target period as an average carbon emission coefficient; t represents a target period, namely, the year, the quarter and the month; p (P) i And R is i Representing the active power and carbon emission factor of the ith power plant; p (P) k And R is k Active power and carbon emission factors of the kth power receiving node respectively; g j Indicating the real-time net power generated by the j-th institution.
As a further improvement of the present technical solution, in the step S3, the carbon emission amount of the public institution is:
wherein C is total carbon emission of public institutions; r is R h A carbon emission factor representing an h-th energy source;indicating the consumption of the h energy by the q-th institution.
As a further improvement of the present technical solution, in the step S4, the LSTM neuron structure is determined, and the model network structure of the LSTM is designed, which specifically includes:
F t representing a forget gate, the forward propagation calculation is:
F t ＝σ(W f [h t-1 ，x t ]+b f )
I t representing the input gate, the forward propagation calculation is:
I t ＝σ(W i [h t-1 ，x t ]+b i )
O t representing the output gate, the forward propagation calculation is:
O t ＝σ(W o [h t-1 ，x t ]+b o )
the memory unit is calculated by the following steps:
in the above, h t-1 Representing the hidden state, which is the output of the last cell of the LSTM; x is x t An output representing the current cell; c t And (3) withCandidate values of the long-term state and the instant state respectively; sigma and tanh represent sigmoid function and activation function, b, respectively m (m=i, f, o, c) is a bias term; w (W) f Weight representing forgetting gate, b f Deviation indicating a forgetful door; w (W) i Representing the weight of the input gate, b i Representing the deviation of the input gate; w (W) o Representing the weight of the output gate, b o Indicating the deviation of the output gate; w (W) c Weights representing candidate values, b c Representing the deviation of the candidate values.
As a further improvement of the present technical solution, in the step S4, the method for initializing the data includes that the input training set and test set data x are converted into a number Z between 0 and 1 by using a MIN-MAX standardization method, and the conversion formula is as follows:
wherein x is i 、x j All belong to the original sequence x 1 ，x 2 ，...，x n ，i∈[1，n]，j∈[1，n]；Z i Belonging to a new sequence Z 1 ，Z 2 ，...，Z n New sequence Z 1 ，Z 2 ，...，Z n ∈[0，1]And is dimensionless.
As a further improvement of the present technical solution, in the step S5, in training the accuracy of the prediction model, the calculation formula of each evaluation index is:
the average absolute percentage error calculation formula is:
the root mean square error calculation formula is:
the deterministic correlation coefficient calculation formula is:
in the above formula, MAPE represents the mean absolute percentage error, RMSE represents the root mean square error, R 2 Representing deterministic correlation coefficients, p i Is an estimated value, q i Is a true value of the code,and->The average of the predicted estimate and the true value, respectively. The lower the MAPE and RMSE values, the more accurate the prediction of the model; r is R 2 The higher the value, the better the prediction effect.
The second object of the present invention is to provide a carbon emission prediction system operation platform device, which includes a processor, a memory, and a computer program stored in the memory and running on the processor, wherein the processor is configured to implement the steps of the above-mentioned LSTM cyclic neural network carbon emission prediction method facing public institutions when executing the computer program.
It is a further object of the present invention to provide a computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the above-described institutional-oriented LSTM recurrent neural network carbon emission prediction method.
Compared with the prior art, the invention has the beneficial effects that:
1. in the public institution-oriented LSTM circulating neural network carbon emission prediction method, carbon emission under the time scale of a public institution is obtained by analyzing the mapping relation between the excavated carbon emission and the energy data and considering the power grid emission coefficient of the area of the public institution according to various energy consumption of the public institution; classifying and preprocessing the carbon emission data of the public institution, and determining an optimal LSTM prediction model to predict the carbon emission of the public institution through multiple training and optimization; the carbon emission sequence classified based on the attribute characteristics of the public institutions is predicted, and carbon reduction requirements and measures can be timely provided for the public institutions according to the predicted results; the carbon emission sequence based on the public institution energy utilization characteristic classification is predicted, and the energy utilization structure of the public institution can be timely adjusted according to the carbon emission trend of various energy sources; the system can provide carbon emission index calculation and management capability for public institutions, and provide effective and high-reliability data support for low-carbon emission of public institutions;
2. in the public institution-oriented LSTM circulating neural network carbon emission prediction method, a carbon emission prediction technology is established for public institutions, future carbon emission trend of public institutions in the smart city construction process is analyzed, important aspects of carbon emission change of the public institutions are analyzed, energy consumption structure and consumption are timely adjusted according to prediction results, an improvement thought is effectively provided for low-carbon construction of the public institutions in time, and accuracy and scientificity of public institution carbon emission prediction are improved.
Drawings
FIG. 1 is a diagram illustrating the operation of the modules of an exemplary LSTM prediction model of the present invention;
FIG. 2 is a schematic diagram of an exemplary LSTM model neuron structure in accordance with the present invention;
FIG. 3 is a diagram of an exemplary LSTM model network architecture in accordance with the present invention;
fig. 4 is a block diagram of an exemplary electronic computer platform device according to the present invention.
Detailed Description
The following description of the embodiments of the present invention 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 invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
1-3, the embodiment provides a public institution-oriented LSTM cyclic neural network carbon emission prediction method, which firstly utilizes the advantage that an LSTM neural network model can process long-time sequences to construct a single-feature input model of long-time scale carbon emission and a multi-feature input model based on carbon emission influence factors; then training the predictive model with the training set and the test set, using MAPE, RMSE, R 2 Evaluating the prediction model, correcting model parameters, and continuously optimizing to obtain an optimal prediction model for predicting the emission of the next periodA quantity fluctuation trend; as shown in fig. 1, the method specifically comprises the following steps.
S1, data acquisition: firstly, determining an public institution to be analyzed, acquiring basic information of the public institution, energy consumption data of the public institution in a certain period and carbon emission data of the public institution in the period, and classifying and processing the carbon emission data;
firstly, determining an public institution to be analyzed, and acquiring basic information of the public institution, wherein the basic information comprises building position, building area, application type and building quantity information so as to determine attribute characteristics of the public institution;
the energy consumption data of the public institution in a certain period (preferably 15 years) comprises coal consumption, oil consumption, gas consumption, electricity consumption and heat consumption data so as to determine the energy consumption characteristics of the public institution and the carbon emission and the electricity generation of a power grid in the area of the public institution;
the carbon emission data is classified, namely, the carbon emission data of the public institutions in the month, the quarter and the year are classified from the two viewpoints of the application attribute of the public institutions and the application energy characteristics of the public institutions.
The electricity consumption condition of the public institution in different time periods can influence the carbon emission accounting, but the carbon emission factors of the power generation stage are difficult to calculate in real time, so that the embodiment collects the annual, quaternary and monthly carbon emission data and electricity consumption data of the regional power grid where the public institution is located, and calculates the annual, quaternary and monthly average carbon emission factors of the public institution.
S2, analysis of total carbon emission: constructing a public institution based on the carbon emission and the generated energy of the power grid in the area of the public institution, and constructing an electric power average carbon emission coefficient calculation model based on a time sequence;
in this step, the expression of the electric power average carbon emission coefficient calculation model based on the time series is:
wherein R is t Is an average carbon emission systemA number indicating carbon emissions generated per unit power generation amount in the target period; t represents a target period, namely, the year, the quarter and the month; p (P) i And R is i Representing the active power and carbon emission factor of the ith power plant; p (P) k And R is k Active power and carbon emission factors of the kth power receiving node respectively; g j Indicating the real-time net power generated by the j-th institution.
S3, carbon emission accounting: firstly, classifying according to attribute characteristics of public institutions, and accounting carbon emission of different types of public institutions; then, aiming at different energy types, consumption and carbon emission factors thereof, calculating the carbon emission of each energy in the calculation period; comprehensively obtaining carbon emission time sequences of public institutions of different types and carbon emission time sequences of public institutions of different energy consumption types;
in this step, the carbon emissions of the public institution are:
wherein C is total carbon emission of public institutions; r is R h A carbon emission factor representing an h-th energy source;indicating the consumption of the h energy by the q-th institution.
S4, determining an LSTM neuron structure, and designing an LSTM model network structure; establishing a structure of an LSTM model, selecting a required loss function, and initializing model parameters; the data input needs to be initialized;
in this step, the LSTM neuron structure is shown in fig. 2, and the model network structure of LSTM is shown in fig. 3, in which: h is a t-1 Representing the hidden state, which is the output of the last cell of the LSTM; x is x t An output representing the current cell; c t And (3) withCandidate values of the long-term state and the instant state respectively; sigma and tanh fractionsRespectively representing a sigmoid function and an activation function, b m (m=i, f, o, c) is a bias term; then there are:
F t representing a forget gate, the forward propagation calculation is:
F t ＝σ(W f [h t-1 ，x t ]+b f )
I t representing the input gate, the forward propagation calculation is:
I t ＝σ(W i [h t-1 ，x t ]+b i )
O t representing the output gate, the forward propagation calculation is:
O t ＝σ(W o [h t-1 ，x t ]+b o )
the memory unit is calculated by the following steps:
in the above, W f Weight representing forgetting gate, b f Deviation indicating a forgetful door; w (W) i Representing the weight of the input gate, b i Representing the deviation of the input gate; w (W) o Representing the weight of the output gate, b o Indicating the deviation of the output gate; w (W) c Weights representing candidate values, b c Representing the deviation of the candidate values.
Further, the method for initializing the data is to convert the inputted training set and testing set data x into a number Z between 0 and 1 by adopting a MIN-MAX standardization method, and the conversion formula is as follows:
wherein x is i 、x j All belong to the original sequence x 1 ，x 2 ，...，x n ，i∈[1，n]，j∈[1，n]；Z i Belonging to a new sequence Z 1 ，Z 2 ，...，Z n New sequence Z 1 ，Z 2 ，...，Z n ∈[0，1]And is dimensionless.
Specifically, the network structure of the LSTM neural network has a 'memory unit', and the memory unit in the hidden layer can be used for screening the information of the last period, so that the LSTM neural network has the capability of processing long-time sequences. The carbon emission of public institutions can be accurately predicted by using the LSTM model.
Further, the time series data of the carbon emission of the public institution are input into the LSTM model respectively, the LSTM model is trained by utilizing the training set and the test set, and the parameters of the model are adjusted through the judgment value of the evaluation index to carry out the next training. In each optimization process, the data processing mode of the embodiment is unchanged, and the data of the training set and the test set adopt consistent time dimension and time division rules. And finally determining the optimal LSTM model through multiple training and adjustment.
S5, training the accuracy of the prediction model: after data input, the estimated value of the model is solved through forward calculation, and the average absolute percentage error (MAPE), the Root Mean Square Error (RMSE) and the deterministic correlation coefficient (R) are selected 2 ) Calculating an estimated value and a true value to measure the effect of model prediction on unknown samples;
in this step, in training the accuracy of the prediction model, the calculation formula of each evaluation index is:
the average absolute percentage error calculation formula is:
the root mean square error calculation formula is:
the deterministic correlation coefficient calculation formula is:
in the above formula, MAPE represents the mean absolute percentage error, RMSE represents the root mean square error, R 2 Representing deterministic correlation coefficients, p i Is an estimated value, q i Is a true value of the code,and->The average of the predicted estimate and the true value, respectively. The lower the MAPE and RMSE values, the more accurate the prediction of the model; r is R 2 The higher the value, the better the prediction effect.
In this embodiment, an average absolute percentage error (MAPE) may be selected as a main evaluation index, and a Root Mean Square Error (RMSE) and an average absolute error (MAE) are adopted to assist in determining the prediction accuracy of the model adopted in the scheme, where the smaller the numbers of the three evaluation indexes, the smaller the difference between the prediction data and the real data, and the higher the accuracy of the prediction model. And evaluating the model precision according to the numerical value of the evaluation index, and sequentially optimizing and adjusting the parameters of the model until an optimal prediction model with high prediction performance is obtained.
S6, determining an optimal prediction model: and continuously circulating the steps S4-S5, optimally adjusting parameters of the LSTM model, establishing a proper LSTM model, and predicting carbon emission of the public institution by using the optimal LSTM model.
Finally, through training and optimizing the LSTM prediction model for multiple times, an optimal LSTM model can be obtained, and further the carbon emission of a public institution in the next period can be predicted by using the LSTM model.
According to the scheme, through data preparation, the application attribute, energy consumption data and other information of the public institution are combined, the carbon emission data of the public institution in different time dimensions are classified and processed, the power grid data of the area where the public institution is located are combined, a power carbon emission coefficient model based on a time sequence is constructed, an LSTM model network structure is designed, the LSTM model is trained and optimized through the data of the public institution, the LSTM network parameters are adjusted to determine an optimal LSTM model, and the carbon emission of the next period is analyzed and predicted.
As shown in fig. 4, the present embodiment also provides a carbon emission prediction system operation platform device, which includes a processor, a memory, and a computer program stored in the memory and running on the processor.
The processor comprises one or more than one processing core, the processor is connected with the memory through a bus, the memory is used for storing program instructions, and the steps of the public institution-oriented LSTM cyclic neural network carbon emission prediction method are realized when the processor executes the program instructions in the memory.
Alternatively, the memory may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.
In addition, the invention also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the steps of the public institution-oriented LSTM circulating neural network carbon emission prediction method when being executed by a processor.
Optionally, the present invention also provides a computer program product containing instructions which, when run on a computer, cause the computer to perform the steps of the above-described aspects of the institutional-oriented LSTM cyclic neural network carbon emission prediction method.
It will be appreciated by those of ordinary skill in the art that the processes for implementing all or part of the steps of the above embodiments may be implemented by hardware, or may be implemented by a program for instructing the relevant hardware, and the program may be stored in a computer readable storage medium, where the above storage medium may be a read-only memory, a magnetic disk or optical disk, etc.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. A public institution-oriented LSTM circulating neural network carbon emission prediction method is characterized in that a single-feature input model of long-time scale carbon emission and a multi-feature input model based on carbon emission influence factors are constructed by utilizing an LSTM neural network model; then training the predictive model with the training set and the test set, using MAPE, RMSE, R 2 Evaluating the prediction model, correcting model parameters, and continuously optimizing to obtain an optimal prediction model to predict the fluctuation trend of the emission in the next period; the method specifically comprises the following steps:
s1, data acquisition: firstly, determining an public institution to be analyzed, acquiring basic information of the public institution, energy consumption data of the public institution in a certain period and carbon emission data of the public institution in the period, and classifying and processing the carbon emission data;
s2, analysis of total carbon emission: constructing a public institution based on the carbon emission and the generated energy of the power grid in the area of the public institution, and constructing an electric power average carbon emission coefficient calculation model based on a time sequence;
s3, carbon emission accounting: firstly, classifying according to attribute characteristics of public institutions, and accounting carbon emission of different types of public institutions; then, aiming at different energy types, consumption and carbon emission factors thereof, calculating the carbon emission of each energy in the calculation period; comprehensively obtaining carbon emission time sequences of public institutions of different types and carbon emission time sequences of public institutions of different energy consumption types;
s4, determining an LSTM neuron structure, and designing an LSTM model network structure; establishing a structure of an LSTM model, selecting a required loss function, and initializing model parameters; the data input needs to be initialized;
s5, training the accuracy of the prediction model: after data input, calculating an estimated value of the model through forward calculation, and calculating the estimated value and a true value by adopting a selected average absolute percentage error, a root mean square error and a deterministic correlation coefficient so as to measure the effect of model prediction on unknown samples;
s6, determining an optimal prediction model: and continuously circulating the steps S4-S5, optimally adjusting parameters of the LSTM model, establishing a proper LSTM model, and predicting carbon emission of the public institution by using the optimal LSTM model.
2. The method for predicting carbon emission of LSTM circulating neural network for public institutions according to claim 1, wherein in the step S1, in the data acquisition, first determining the public institutions to be analyzed, and obtaining basic information of the public institutions, including building position, building area, usage type, and building number information, to determine attribute characteristics of the public institutions;
the energy consumption data of the public institution comprise coal consumption, oil consumption, gas consumption, electricity consumption and heat consumption data in a certain period so as to determine the energy consumption characteristics of the public institution, and the carbon emission and the electricity generation of the power grid of the area where the public institution is located;
the carbon emission data is classified, namely, the carbon emission data of the public institutions in the month, the quarter and the year are classified from the two viewpoints of the application attribute of the public institutions and the application energy characteristics of the public institutions.
3. The method for predicting carbon emission of LSTM cyclic neural network for public institution according to claim 1, wherein in step S2, the expression of the calculation model of average carbon emission coefficient of electric power based on time sequence is:
wherein R is t Representing carbon emission generated per unit power generation amount in a target period as an average carbon emission coefficient; t represents a target period, namely, the year, the quarter and the month; p (P) i And R is i Represents the ithActive power and carbon emission factor of the individual power plants; p (P) k And R is k Active power and carbon emission factors of the kth power receiving node respectively; g j Indicating the real-time net power generated by the j-th institution.
4. The method for predicting carbon emissions of an LSTM circulating neural network for public institutions according to claim 1, wherein in the step S3, the carbon emissions of the public institutions are as follows:
wherein C is total carbon emission of public institutions; r is R h A carbon emission factor representing an h-th energy source;indicating the consumption of the h energy by the q-th institution.
5. The public institution-oriented LSTM circulating neural network carbon emission prediction method according to claim 1, wherein in the step S4, an LSTM neuron structure is determined, and a model network structure of LSTM is designed, specifically:
F t representing a forget gate, the forward propagation calculation is:
F t ＝σ(W f [h t-1 ，x t ]+b f )
I t representing the input gate, the forward propagation calculation is:
I t ＝σ(W i [h t-1 ，x t ]+b i )
O t representing the output gate, the forward propagation calculation is:
O t ＝σ(W o [h t-1 ，x t ]+b o )
the memory unit is calculated by the following steps:
in the above, h t-1 Representing the hidden state, which is the output of the last cell of the LSTM; x is x t An output representing the current cell; c t And (3) withCandidate values of the long-term state and the instant state respectively; sigma and tanh represent sigmoid function and activation function, b, respectively m (m=i, f, o, c) is a bias term; w (W) f Weight representing forgetting gate, b f Deviation indicating a forgetful door; w (W) i Representing the weight of the input gate, b i Representing the deviation of the input gate; w (W) o Representing the weight of the output gate, b o Indicating the deviation of the output gate; w (W) c Weights representing candidate values, b c Representing the deviation of the candidate values.
6. The method for predicting carbon emission of LSTM cyclic neural network for public institution according to claim 5, wherein in step S4, the method for initializing data is to convert the inputted training set and test set data x into a number Z between 0 and 1 by using MIN-MAX standardization method, and the conversion formula is as follows:
wherein x is i 、x j All belong to the original sequence x 1 ，x 2 ，...，x n ，i∈[1，n]，j∈[1，n]；Z i Belonging to a new sequence Z 1 ，Z 2 ，...，Z n New sequence Z 1 ，Z 2 ，...，Z n ∈[0，1]And is dimensionless.
7. The method for predicting carbon emission of LSTM cyclic neural network for public institution according to claim 1, wherein in step S5, in training the accuracy of the prediction model, the calculation formula of each evaluation index is:
the average absolute percentage error calculation formula is:
the root mean square error calculation formula is:
the deterministic correlation coefficient calculation formula is:
in the above formula, MAPE represents the mean absolute percentage error, RMSE represents the root mean square error, R 2 Representing deterministic correlation coefficients, p i Is an estimated value, q i Is a true value of the code,and->The average of the predicted estimate and the true value, respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310893182.9A CN116882580A (en) | 2023-07-20 | 2023-07-20 | Public institution-oriented LSTM circulating neural network carbon emission prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310893182.9A CN116882580A (en) | 2023-07-20 | 2023-07-20 | Public institution-oriented LSTM circulating neural network carbon emission prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116882580A true CN116882580A (en) | 2023-10-13 |
Family
ID=88256555
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310893182.9A Pending CN116882580A (en) | 2023-07-20 | 2023-07-20 | Public institution-oriented LSTM circulating neural network carbon emission prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116882580A (en) |
-
2023
- 2023-07-20 CN CN202310893182.9A patent/CN116882580A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113572206B (en) | Wind power output interval prediction method | |
| CN113064093B (en) | Method and system for jointly estimating state of charge and state of health of energy storage battery | |
| CN109993270A (en) | Lithium ion battery residual life prediction technique based on grey wolf pack optimization LSTM network | |
| CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN111693931A (en) | Intelligent electric energy meter error remote calculation method and device and computer equipment | |
| CN115099511A (en) | Photovoltaic power probability estimation method and system based on optimized copula | |
| Zou et al. | Wind turbine power curve modeling using an asymmetric error characteristic-based loss function and a hybrid intelligent optimizer | |
| CN112630659A (en) | Lithium battery SOC estimation method based on improved BP-EKF algorithm | |
| CN112834927A (en) | Lithium battery residual life prediction method, system, device and medium | |
| CN111008790A (en) | Hydropower station group power generation electric scheduling rule extraction method | |
| CN110807508B (en) | Bus peak load prediction method considering complex weather influence | |
| Cruz et al. | Neural network prediction interval based on joint supervision | |
| Mohammad et al. | Short term load forecasting using deep neural networks | |
| CN114897264A (en) | Photovoltaic output interval prediction method under small sample scene based on transfer learning | |
| CN117686937A (en) | Method for estimating health state of single battery in battery system | |
| CN117291069A (en) | LSTM sewage water quality prediction method based on improved DE and attention mechanism | |
| CN116882580A (en) | Public institution-oriented LSTM circulating neural network carbon emission prediction method | |
| CN114861555A (en) | Regional comprehensive energy system short-term load prediction method based on Copula theory | |
| CN115310355A (en) | Multi-energy coupling-considered multi-load prediction method and system for comprehensive energy system | |
| CN112581311B (en) | Method and system for predicting long-term output fluctuation characteristics of aggregated multiple wind power plants | |
| CN112685900B (en) | Power load simulation method for representing impact load power characteristics | |
| Shen et al. | An interval analysis scheme based on empirical error and MCMC to quantify uncertainty of wind speed | |
| Wang et al. | An alternative method for estimating wind-power capacity credit based on reliability evaluation using intelligent search | |
| Mbasso et al. | Demand-supply forecasting based on deep learning for electricity balance in Cameroon |
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 |