Carbon emission prediction system and method for coal-fired power plant based on RBF neural network

Abstract

The invention relates to the technical field of coal-fired power plants, and discloses a carbon emission prediction system for a coal-fired power plant based on an RBF neural network, and the system comprises a data obtaining module which is used for obtaining electric power data, fire coal data and working condition data related to the coal-fired power plant; the screening module is used for analyzing the data acquired by the data acquisition module and the carbon emission of the coal-fired power plant based on a Pearson's correlation method and screening out variable data with large correlation; and the prediction module is used for predicting the carbon emission by the RBF neural network model according to the input real-time variable data. According to the carbon emission prediction system and method for the coal-fired power plant based on the RBF neural network, the power plant influence variable with large correlation is selected as the input variable through Pearson analysis, the carbon emission is predicted through the RBF neural network, and the system and method have the advantages that the accuracy of the prediction result is improved, the prediction deviation is reduced, and the convergence rate and precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal-fired power plants, and in particular to a carbon emission prediction system and method for coal-fired power plants based on RBF neural network. Background Art

[0002] In recent years, global environmental issues have received more and more attention. One of the issues that cannot be ignored is global warming. Excessive emissions of greenhouse gases such as CO2 are one of the main factors causing global warming and climate change. According to data released by the IPCC, stationary source combustion currently causes about 70% of greenhouse gas emissions in the energy sector, of which coal-fired power plants account for half of the emissions, making them one of the main sources of carbon dioxide emissions.

[0003] With the advancement of energy conservation and emission reduction in the whole society, the problem of carbon emission reduction methods and control of carbon emission peak of coal-fired power plants needs to be solved urgently. The key data is the acquisition of carbon emissions. In the past, the calculation of carbon emissions of power plants was mainly based on the "2006 IPCC National Greenhouse Gas Guidelines", which mainly considered the impact of relevant factors in the production process of power plants on CO2 emissions, so as to calculate and control carbon emissions. However, this method is usually in a passive management state and is prone to ignore the impact of other factors, making it impossible to carry out targeted and proactive prevention of carbon emissions.

[0004] Therefore, it is necessary to predict the carbon emissions of coal-fired power plants. Currently, there are few studies on methods for predicting carbon emissions from coal-fired power plants, and traditional prediction models (such as artificial neural networks, support vector machines, etc.) are mainly used for prediction. Although traditional prediction models are simple, they have the disadvantages of large errors in prediction results, inaccurate predictions, and difficulty in model convergence.

[0005] Therefore, a carbon emission prediction system and method for coal-fired power plants based on RBF neural network is proposed to solve the above problems. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the shortcomings of the prior art, the present invention provides a carbon emission prediction system and method for coal-fired power plants based on RBF neural network, which solves the problems of large prediction result errors, inaccurate predictions, and difficult model convergence in the existing traditional prediction models for coal-fired power plants (such as artificial neural networks, support vector machines, etc.).

[0008] (II) Technical solution

[0009] The technical solution of the present invention to solve the above technical problems is as follows: A carbon emission prediction system for coal-fired power plants based on RBF neural network, comprising:

[0010] A data acquisition module is used to acquire power data, coal combustion data and operating condition data related to the coal-fired power plant;

[0011] A screening module, used to analyze the data acquired by the data acquisition module and the carbon emissions of coal-fired power plants based on the Pearson correlation method, and screen out variable data with high correlation;

[0012] The prediction module uses the RBF neural network model to predict carbon emissions based on the input real-time variable data.

[0013] The beneficial effects of the present invention are: selecting power plant influencing variables with high correlation as input variables through Pearson analysis, and predicting carbon emissions through RBF neural network, which has the characteristics of improving the accuracy of prediction results, reducing prediction deviation, and improving convergence rate and precision.

[0014] Based on the above technical solution, the present invention can also be improved as follows.

[0015] Furthermore, it also includes:

[0016] Training module: Construct and train the optimal RBF neural network model based on the variable data screened by the screening module input.

[0017] A method for predicting carbon emissions from coal-fired power plants based on RBF neural networks comprises the following steps:

[0018] 1) Obtain the electricity data, coal data and technical equipment data of coal-fired power plants, and analyze them with the carbon emissions of coal-fired power plants based on the Pearson correlation method to screen out variable data with high correlation;

[0019] 2) normalizing the acquired variable data, and constructing and training an RBF neural network model based on the normalized data;

[0020] 3) Predict the carbon emissions of coal-fired power plants based on the trained RBF neural network model, output the carbon emissions prediction results, and complete the prediction.

[0021] Furthermore, the specific process of analyzing the acquired coal-fired power plant data and the carbon emissions of the coal-fired power plant based on the Pearson correlation method in step 1) is as follows:

[0022] The Pearson correlation coefficient is calculated as follows:

[0023]

[0024] Among them, E is the mathematical expectation, cov represents the covariance, X is the variable data of the coal-fired power plant, and Y is the carbon dioxide emission data.

[0025] Furthermore, the specific process of normalizing the variable data obtained in step 2) is as follows:

[0026] Linearly transform the original data and map the data values ​​to [0,1];

[0027] The formula is:

[0028]

[0029] Further, the specific process of constructing and training the RBF neural network model in step 2) is as follows:

[0030] S1. Construct RBF neural network model, and the kernel function is Gaussian kernel function;

[0031] S2, randomly divide the normalized data into training set and test set, and input a part of the training set data into the constructed RBF neural network model;

[0032] S3. Data processing training is performed using the RBF neural network model. If the output result meets the requirements, the training is completed and a trained RBF neural network model is obtained to complete the prediction.

[0033] Furthermore, the data processing training result by the RBF neural network model in step S3 also includes:

[0034] If the output result does not meet the requirements, the parameters to be optimized of the RBF neural network prediction model are updated, and the training set data is continuously input into the RBF neural network prediction model for training until the output result meets the requirements, and the training is terminated.

[0035] Furthermore, the parameters to be optimized of the RBF neural network prediction model are the penalty parameter C and the kernel parameter δ;

[0036] The improved particle swarm algorithm is used to update the penalty parameter C and the kernel parameter δ of the RBF neural network prediction model to be optimized.

[0037] Furthermore, the particle swarm algorithm specifically includes:

[0038] 1) Initialize acceleration constants c1 and c2, maximum number of iterations, and number of particle swarms;

[0039] 2) The improved particle swarm algorithm is:

[0040]

[0041]

[0042] in:

[0043] d=1,2,...,n, n represents the dimension of the influencing factor data;

[0044] i=1,2,...,m, where m represents the size of the particle swarm;

[0045] t represents the number of iterations of the current particle, and the random number of r1 and r2∈[0,1];

[0046] c1 and c2 are acceleration constants;

[0047] t is the number of iterations;

[0048] represents the velocity of the dth dimension of the ith particle in the data feature space at the tth iteration;

[0049] represents the velocity of the dth dimension of the ith particle in the data feature space at the (t+1)th iteration;

[0050] represents the coordinate of the d-th dimension of the i-th particle in the feature space;

[0051] represents the velocity of the dth dimension of the ith particle in the data feature space at the (t-1)th iteration;

[0052] It means that the speed of particle iteration update is related to the number of iterations and the previous speed, so that the optimal value can be found faster to shorten the convergence time;

[0053] represents the coordinate of the d-th dimension of the i-th particle in the feature space;

[0054] represents the individual optimal solution of the particle;

[0055] Indicates the global optimal solution of the particle swarm, maintaining the speed and direction of the previous iteration;

[0056] For self-awareness, It is a global cognitive item;

[0057] 3) Use step 2) to obtain new penalty parameter C and kernel parameter σ, and update them to the RBF neural network prediction model.

[0058] Furthermore, the process of obtaining the carbon emission prediction model includes:

[0059] 1) There are P input samples X P , the response target in the output space is y p , the nonlinear mapping F(X) is obtained as follows:

[0060] F(X p )=y p ;

[0061] Select P basis functions, each basis function corresponds to a training data, and the form of each basis function is as follows:

[0062] g(||XX p ||),p=1,2,......p;

[0063] in:

[0064] X P is the center of the function;

[0065] g is the point X and the center X of the input space p The distance is the independent variable;

[0066] Define F(X) as a linear combination of basis functions, let ω p is the weight between the pth hidden units, then:

[0067]

[0068]

[0069]

[0070]

[0071] Let g ip =g(||XX p ||), i=1,2,……p, the above equations can be rewritten as:

[0072]

[0073] Use G to represent g ip For a P×P matrix, Gω=y. If G is reversible, we can solve ω=G -1 y;

[0074] Also includes:

[0075]

[0076] Where x is the input vector of the sample object;

[0077] k j (x) is the output vector of the i-th hidden layer node;

[0078] c i is the center vector of the i-th hidden layer node;

[0079] δ i is the normalization constant of the i-th hidden layer node;

[0080] get:

[0081] T=ω·k j (χ);

[0082] T is the carbon emission prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is the overall flow chart of the present invention;

[0084] Figure 2 is a flow chart of the method of the present invention;

[0085] Figure 3 This is a system module diagram of the present invention. DETAILED DESCRIPTION

[0086] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0087] In the embodiment, Figure 1 , Figure 2 and Figure 3 A carbon emission prediction system for coal-fired power plants based on RBF neural network is given, including:

[0088] A data acquisition module is used to acquire power data, coal combustion data and operating condition data related to the coal-fired power plant;

[0089] Specifically, the data obtained include variable data such as power generation of power plants, power supply, coal consumption, carbon content of received basic elements, carbon content per unit calorific value, and carbon emission intensity of power supply;

[0090] A screening module is used to analyze the data acquired by the data acquisition module and the carbon emissions of coal-fired power plants based on the Pearson correlation method to screen out variable data with high correlation;

[0091] Furthermore, it also includes:

[0092] Training module: construct and train the optimal RBF neural network model based on the variable data screened by the screening module input;

[0093] The prediction module uses the RBF neural network model to predict carbon emissions based on the input real-time variable data;

[0094] Among them, the kernel function of the RBF neural network model is a Gaussian kernel function.

[0095] A method for predicting carbon emissions from coal-fired power plants based on RBF neural networks comprises the following steps:

[0096] 1) Obtain the electricity data, coal data and technical equipment data of coal-fired power plants, and analyze them with the carbon emissions of coal-fired power plants based on the Pearson correlation method to screen out variable data with high correlation;

[0097] Furthermore, in step 1), the specific process of analyzing the acquired coal-fired power plant data and the carbon emissions of the coal-fired power plants based on the Pearson correlation method is as follows:

[0098] The Pearson correlation coefficient is calculated as follows:

[0099]

[0100] Among them, E is the mathematical expectation, cov represents the covariance, X is the variable data of the coal-fired power plant, and Y is the carbon dioxide emission data;

[0101] Coal-fired power plant data include historical electricity data of coal-fired power plants, power generation, power supply, coal consumption, received basic element carbon content, unit calorific value carbon content, power supply carbon emission intensity, etc.;

[0102] 2) Normalizing the acquired variable data, and constructing and training an RBF neural network model based on the normalized data;

[0103] Furthermore, the specific process of normalizing the acquired variable data in step 2) is as follows:

[0104] Linearly transform the original data and map the data values ​​to [0,1];

[0105] The formula is:

[0106]

[0107] Further, the specific process of constructing and training the RBF neural network model in step 2) is as follows:

[0108] S1. Construct RBF neural network model, and the kernel function is Gaussian kernel function;

[0109] S2, randomly dividing the normalized data into a training set and a test set, preferably, dividing them into 7:3, and inputting a part of the training set data into the constructed RBF neural network model;

[0110] S3, data processing training is performed through the RBF neural network model. If the output result meets the requirements, the training is completed, and the trained RBF neural network model is obtained to complete the prediction;

[0111] Furthermore, the data processing training result by the RBF neural network model in step S3 also includes:

[0112] If the output result does not meet the requirements, the parameters to be optimized of the RBF neural network prediction model are updated, and the training set data is continuously input into the RBF neural network prediction model for training until the output result meets the requirements, and the training is terminated.

[0113] Furthermore, the parameters to be optimized of the RBF neural network prediction model are the penalty parameter C and the kernel parameter δ;

[0114] Among them, the improved particle swarm algorithm is used to update the penalty parameter C and the kernel parameter δ of the RBF neural network prediction model.

[0115] Furthermore, the particle swarm algorithm specifically includes:

[0116] 1) Initialize acceleration constants c1 and c2, maximum number of iterations, and number of particle swarms;

[0117] 2) The improved particle swarm algorithm is:

[0118]

[0119]

[0120] in:

[0121] d=1,2,...,n, n represents the dimension of the influencing factor data;

[0122] i=1,2,...,m, where m represents the size of the particle swarm;

[0123] t represents the number of iterations of the current particle, and the random number of r1 and r2∈[0,1];

[0124] c1 and c2 are acceleration constants;

[0125] t is the number of iterations;

[0126] represents the velocity of the dth dimension of the ith particle in the data feature space at the tth iteration;

[0127] represents the velocity of the dth dimension of the ith particle in the data feature space at the (t+1)th iteration;

[0128] represents the coordinate of the d-th dimension of the i-th particle in the feature space;

[0129] represents the velocity of the dth dimension of the ith particle in the data feature space at the (t-1)th iteration;

[0130] It means that the speed of particle iteration update is related to the number of iterations and the previous speed, so that the optimal value can be found faster to shorten the convergence time;

[0131] represents the coordinate of the d-th dimension of the i-th particle in the feature space;

[0132] represents the individual optimal solution of the particle;

[0133] Indicates the global optimal solution of the particle swarm, maintaining the speed and direction of the previous iteration;

[0134] For self-awareness, It is a global cognitive item;

[0135] 3) Use step 2) to obtain new penalty parameter C and kernel parameter σ, and update them to the RBF neural network prediction model.

[0136] 3) Predict the carbon emissions of coal-fired power plants based on the trained RBF neural network model, output the carbon emissions prediction results, and complete the prediction.

[0137] RBF neural network model: the input data is directly mapped to the hidden layer space, and the radial basis function is used as the "basis" of the hidden unit to form the hidden layer space. The input data is transformed in this space, and the nonlinear data in the low-dimensional space is transformed into linearly separable data in the high-dimensional space. This nonlinear mapping relationship is determined by the center point of the radial basis function.

[0138] The output layer is obtained through the linear mapping of the hidden layer, that is, the output of the network is the linear weighted sum of the outputs of the hidden layer neural units.

[0139] The hidden layer is the most important layer in the RBF neural network model. The number of its units depends on the specific situation of the problem to be solved. The radial basis function of the hidden layer is a scalar function that is symmetrical along the radial direction. It is the distance from any point X in space to the center of a certain point c. i Single point function of the Euclidean distance between;

[0140] Furthermore, the process of obtaining the carbon emission prediction model includes:

[0141] 1) There are P input samples X P, the response target in the output space is y p , the nonlinear mapping F(X) is obtained as follows:

[0142] F(X p )=y p ;

[0143] Select P basis functions, each basis function corresponds to a training data, and the form of each basis function is as follows:

[0144] g(||XX p ||),p=1,2,......p;

[0145] in:

[0146] X P is the center of the function;

[0147] g is the point X and the center X of the input space p The distance is the independent variable;

[0148] Define F(X) as a linear combination of basis functions, let ω p is the weight between the pth hidden units, then:

[0149]

[0150]

[0151]

[0152]

[0153] Let g ip =g(||XX p ||), i=1,2,……p, the above equations can be rewritten as:

[0154]

[0155] Use G to represent g ip For a P×P matrix, Gω=y. If G is reversible, we can solve ω=G -1 y;

[0156] Also includes:

[0157]

[0158] Where x is the input vector of the sample object;

[0159] k j (x) is the output vector of the i-th hidden layer node;

[0160] c iis the center vector of the i-th hidden layer node;

[0161] δ i is the normalization constant of the i-th hidden layer node, and is also the width parameter of the center vector kernel function, i.e., the variance (spread constant);

[0162] get:

[0163] T=ω·k j (χ);

[0164] T is the carbon emission prediction model

[0165] The above prediction method can quickly and accurately obtain the prediction results of carbon emissions based on the variable input, and the prediction results are more accurate. The advance prediction data can be used to solve the problems of carbon emission reduction methods and controlling carbon emission peaks of power plants.

[0166] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0167] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A carbon emission prediction system for coal-fired power plants based on RBF neural network, characterized by: include: A data acquisition module is used to acquire power data, coal combustion data and operating condition data related to the coal-fired power plant; A screening module, used to analyze the data acquired by the data acquisition module and the carbon emissions of coal-fired power plants based on the Pearson correlation method, and screen out variable data with high correlation; The prediction module uses the RBF neural network model to predict carbon emissions based on the input real-time variable data.

2. A carbon emission prediction system for coal-fired power plants based on RBF neural network according to claim 1, characterized in that: Also includes: Training module: Construct and train the optimal RBF neural network model based on the variable data screened by the screening module input.

3. A method for predicting carbon emissions from coal-fired power plants based on RBF neural network, characterized in that: The steps include: 1) Obtain the electricity data, coal data and technical equipment data of coal-fired power plants, and analyze them with the carbon emissions of coal-fired power plants based on the Pearson correlation method to screen out variable data with high correlation; 2) normalizing the acquired variable data, and constructing and training an RBF neural network model based on the normalized data; 3) Predict the carbon emissions of coal-fired power plants based on the trained RBF neural network model, output the carbon emissions prediction results, and complete the prediction.

4. The carbon emission prediction method for coal-fired power plants based on RBF neural network according to claim 3 is characterized in that: The specific process of analyzing the acquired coal-fired power plant data and the carbon emissions of the coal-fired power plant based on the Pearson correlation method in step 1) is as follows: The Pearson correlation coefficient is calculated as follows: Among them, E is the mathematical expectation, cov represents the covariance, X is the variable data of the coal-fired power plant, and Y is the carbon dioxide emission data.

5. The carbon emission prediction method for coal-fired power plants based on RBF neural network according to claim 3 is characterized in that: The specific process of normalizing the variable data obtained in step 2) is as follows: Linearly transform the original data and map the data values ​​to [0,1]; The formula is:

6. The carbon emission prediction method for coal-fired power plants based on RBF neural network according to claim 3 is characterized in that: The specific process of constructing and training the RBF neural network model in step 2) is as follows: S1. Construct RBF neural network model, and the kernel function is Gaussian kernel function; S2, randomly divide the normalized data into training set and test set, and input a part of the training set data into the constructed RBF neural network model; S3. Data processing training is performed using the RBF neural network model. If the output result meets the requirements, the training is completed and a trained RBF neural network model is obtained to complete the prediction.

7. A method for predicting carbon emissions from coal-fired power plants based on RBF neural network according to claim 6, characterized in that: The data processing training results of the RBF neural network model in step S3 also include: If the output result does not meet the requirements, the parameters to be optimized of the RBF neural network prediction model are updated, and the training set data is continuously input into the RBF neural network prediction model for training until the output result meets the requirements, and the training is terminated.

8. The method for predicting carbon emissions from coal-fired power plants based on RBF neural network according to claim 7 is characterized by: The parameters to be optimized of the RBF neural network prediction model are the penalty parameter C and the kernel parameter δ; The improved particle swarm algorithm is used to update the penalty parameter C and the kernel parameter δ of the RBF neural network prediction model to be optimized.

9. A method for predicting carbon emissions from coal-fired power plants based on RBF neural network according to claim 8, characterized in that: The particle swarm algorithm specifically includes: 1) Initialize acceleration constants c1 and c2, maximum number of iterations, and number of particle swarms; 2) The improved particle swarm algorithm is: in: d=1,2,...,n, n represents the dimension of the influencing factor data; i=1,2,...,m, where m represents the size of the particle swarm; t represents the number of iterations of the current particle, and the random number of r1 and r2∈[0,1]; c1 and c2 are acceleration constants; t is the number of iterations; represents the velocity of the dth dimension of the ith particle in the data feature space at the tth iteration; represents the velocity of the dth dimension of the ith particle in the data feature space at the (t+1)th iteration; represents the coordinate of the d-th dimension of the i-th particle in the feature space; represents the velocity of the dth dimension of the ith particle in the data feature space at the (t-1)th iteration; Indicates that the speed of particle iterative update is related to the number of iterations and the previous speed; represents the coordinate of the d-th dimension of the i-th particle in the feature space; represents the individual optimal solution of the particle; Indicates the global optimal solution of the particle swarm, maintaining the speed and direction of the previous iteration; For self-awareness, It is a global cognitive item; 3) Use step 2) to obtain new penalty parameter C and kernel parameter σ, and update them to the RBF neural network prediction model.

10. The method for predicting carbon emissions from coal-fired power plants based on RBF neural network according to claim 3 is characterized by: The process of obtaining the carbon emission prediction model includes: 1) There are P input samples X P , the response target in the output space is y p , the nonlinear mapping F(X) is obtained as follows: F(X p )=y p ; Select P basis functions, each basis function corresponds to a training data, and the form of each basis function is as follows: g(||X-X p ||),p=1,2,......p; in: X P is the center of the function; g is the point X and the center X of the input space p The distance is the independent variable; Define F(X) as a linear combination of basis functions, let ω p is the weight between the pth hidden units, then: Let g ip =g(||XX p ||), i=1,2,……p, the above equations can be rewritten as: Use G to represent g ip For a P×P matrix, Gω=y. If G is reversible, we can solve ω=G -1 y; Also includes: Where x is the input vector of the sample object; k j (x) is the output vector of the i-th hidden layer node; c i is the center vector of the i-th hidden layer node; δ i is the normalization constant of the i-th hidden layer node.

Images