Boiler combustion thermal efficiency prediction method and device

Abstract

The present disclosure provides a method and device for predicting boiler combustion thermal efficiency. The method includes: collecting real-time operation data of the boiler, and preprocessing the real-time operation data; dividing the preprocessed real-time operation data into a training set and a test set; clustering the real-time operation data in the training set using an improved K-means algorithm, specifically including: first optimizing the initial clustering center of K-means using a biogeographic optimization algorithm, and then clustering the real-time operation data using a K-means algorithm; setting relevant parameters so that the network conforms to the boiler combustion model; based on the clustered real-time operation data, training a preset extreme learning machine network to obtain the boiler combustion model; and predicting the boiler combustion thermal efficiency using the boiler combustion model. It can effectively improve the prediction accuracy, show its effectiveness, and can effectively improve the prediction accuracy and prediction efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of boiler combustion, and in particular relates to a method and device for predicting boiler combustion thermal efficiency. Background Art

[0002] With the rapid development of my country's economy, people are using more and more energy, among which the use of electricity is increasing day by day. Electricity mainly comes from thermal power generation, hydropower generation, wind power generation, solar power generation and nuclear power generation. Among all power generation methods, thermal power generation has the advantages of low cost, sufficient power and low environmental requirements. These advantages determine that my country's power generation method is still dominated by thermal power generation, and thermal power generation accounts for about 80% of the total power generation. In my country, there are two major problems to be solved for thermal power generation: one is to improve the operating efficiency of the unit and reduce the coal consumption of the unit, and the other is to effectively reduce the emission of harmful gases during the coal burning process. The first task to achieve boiler combustion optimization is to establish a model of the parameters to be optimized. In the process of boiler modeling, artificial neural network modeling can effectively avoid some thermal problems. Artificial neural network technology can establish the model characteristics of the parameters to be optimized based on the historical data of the boiler combustion process without understanding the relevant knowledge of the boiler combustion system. Summary of the invention

[0003] The present disclosure aims to solve at least one of the technical problems existing in the prior art and provide a method and device for predicting boiler combustion thermal efficiency.

[0004] In one aspect of the present disclosure, a method for predicting boiler combustion thermal efficiency is provided, the method comprising:

[0005] Collecting real-time operation data of the boiler and preprocessing the real-time operation data;

[0006] Divide the preprocessed real-time running data into a training set and a test set;

[0007] The improved K-means algorithm is used to cluster the real-time operation data in the training set, specifically comprising: firstly optimizing the initial clustering center of K-means by using a biogeographic optimization algorithm, and then clustering the real-time operation data by using the K-means algorithm;

[0008] Set relevant parameters to make the network conform to the boiler combustion model;

[0009] Based on the clustered real-time operation data, a preset extreme learning machine network is trained to obtain the boiler combustion model;

[0010] The boiler combustion model is used to predict the boiler combustion thermal efficiency.

[0011] In some embodiments, the method of optimizing the initial K-means clustering center using a biogeographic optimization algorithm comprises:

[0012] Initialize the parameters related to the biogeography optimization algorithm;

[0013] Calculate the fitness value of each solution, the number of species, immigration rate and emigration rate, and save individuals with high fitness.

[0014] Use the immigration rate to select the individual variables that need to be migrated in, and use the out-migration rate to select the individual variables that need to be migrated out, and perform the migration operation;

[0015] Perform mutation operations based on mutation probability;

[0016] Whether the iteration stop condition is met, if not, continue to execute the above steps;

[0017] The individuals with high fitness are clustered as the initial center points.

[0018] In some embodiments, the parameters in the biogeographic optimization algorithm are set as follows:

[0019] The maximum number of iterations Maxgen = 200; the maximum population size Popsize = 120; the population size Numvar = 6; the population migration probability Pmodify = 1; the mutation probability Pmutate = 0.01.

[0020] In some embodiments, before training the extreme learning machine network, the method further comprises:

[0021] Based on the improved K-means algorithm, the network structure and node parameters of the extreme learning machine network are optimized and selected.

[0022] In some embodiments, the network structure optimization includes selecting network input variables, the number of network hidden layer nodes, and activation functions;

[0023] The node parameter optimization includes the network hidden layer node parameters w i and b i Optimization.

[0024] In some embodiments, preprocessing the real-time operation data includes:

[0025] The real-time operation data is processed in advance by unweighted group average method UPGMA to remove noise data and edge data.

[0026] Another aspect of the present disclosure provides a boiler combustion thermal efficiency prediction device, the device comprising:

[0027] A collection module, used for collecting real-time operation data of the boiler and preprocessing the real-time operation data;

[0028] A partitioning module is used to partition the preprocessed real-time running data into a training set and a test set;

[0029] A clustering module is used to cluster the real-time operation data in the training set using an improved K-means algorithm, specifically comprising: firstly optimizing the initial K-means clustering center using a biogeographic optimization algorithm, and then clustering the real-time operation data using the K-means algorithm;

[0030] A setting module, used to set relevant parameters so that the network conforms to the boiler combustion model;

[0031] A training module, used for training a preset extreme learning machine network based on the clustered real-time operation data to obtain the boiler combustion model;

[0032] The prediction module is used to predict the boiler combustion thermal efficiency by using the boiler combustion model.

[0033] In some implementations, the clustering module is further configured to:

[0034] Initialize the parameters related to the biogeography optimization algorithm;

[0035] Calculate the fitness value of each solution, the number of species, immigration rate and emigration rate, and save individuals with high fitness.

[0036] Use the immigration rate to select the individual variables that need to be migrated in, and use the out-migration rate to select the individual variables that need to be migrated out, and perform the migration operation;

[0037] Perform mutation operations based on mutation probability;

[0038] Whether the iteration stop condition is met, if not, continue to execute the above steps;

[0039] The individuals with high fitness are clustered as the initial center points.

[0040] Another aspect of the present disclosure provides an electronic device, including:

[0041] one or more processors;

[0042] A storage unit is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors can implement the method described above.

[0043] Another aspect of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program can implement the method described above when executed by a processor.

[0044] The boiler combustion thermal efficiency prediction method and device disclosed in the present invention combine the biogeographic optimization algorithm with the K-means clustering algorithm, aiming to search for the initial cluster center, so that the clustering algorithm has good classification accuracy and global search ability. On the basis of the ELM prediction model, the KBBO optimization algorithm is used for the optimization selection of the network input variables, the configuration and parameters of the hidden layer nodes, and the Tikhonov regularization parameters of the ELM. The method can effectively improve the prediction accuracy and show its effectiveness. It can effectively improve the prediction accuracy and prediction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a method for predicting boiler combustion thermal efficiency according to an embodiment of the present disclosure;

[0046] Figure 2 A structural diagram of a circulating fluidized bed boiler according to another embodiment of the present disclosure;

[0047] Figure 3 is a training error convergence curve of another embodiment of the present disclosure;

[0048] Figure 4 It is a schematic structural diagram of a boiler combustion thermal efficiency prediction device according to another embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, the present disclosure is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0050] In one aspect of this embodiment, Figure 1 As shown, it relates to a boiler combustion thermal efficiency prediction method S100, and the method S100 includes:

[0051] S110, collecting real-time operation data of the boiler, and preprocessing the real-time operation data.

[0052] Specifically, in this step, 600 sets of real-time boiler operation data are collected from a thermal power plant, including 200 sets of data at 50%, 75%, and 100% load. Figure 2The factors that affect thermal efficiency mainly include 20 parameters such as boiler load, bed temperature, primary and secondary air volume, oxygen content in the air, primary and secondary air temperature, etc. All data are pre-processed by the unweighted group average method UPGMA to eliminate noise data and edge data, minimize measurement errors, and ensure the authenticity of experimental data. Among them, the thermal efficiency calculation uses the counter-balance calculation method to calculate the physical heat loss of flue gas, the heat loss of incomplete combustion of solids, and the heat loss of incomplete combustion of gases in the total heat loss.

[0053] S120: Divide the preprocessed real-time operation data into a training set and a test set.

[0054] Specifically, in this step, the 600 groups of samples in the above step S110 are divided into 7:3 ratios, 420 groups are used as training samples, and the remaining 180 groups are used as test samples.

[0055] S130, clustering the real-time operation data in the training set using an improved K-means algorithm, specifically comprising: first optimizing the K-means initial clustering center using a biogeographic optimization algorithm, and then clustering the real-time operation data using the K-means algorithm.

[0056] Specifically, in this step, the steps of optimizing the initial K-means clustering center using the biogeography optimization algorithm specifically include:

[0057] (1) Initialize the relevant parameters of the biogeographic optimization algorithm;

[0058] (2) Calculate the fitness value of each solution, the number of species corresponding to it, and the immigration rate and emigration rate, while preserving individuals with high fitness;

[0059] (3) Use the migration rate to select the individual variables that need to be migrated in, and use the migration rate to select the individual variables that need to be migrated out, and perform the migration operation;

[0060] (4) Perform mutation operation according to mutation probability;

[0061] (5) Whether the iteration stop condition is met, if not, continue to execute the above step (2);

[0062] (6) The individuals with high fitness are clustered as the initial center points.

[0063] After the grouping is completed, clustering is performed to find the sample data with the highest thermal efficiency in each class, and then the 180 test data sets are divided into corresponding classes. Then the data in each class is optimized one by one using the data processing method. Among them, the parameters in the biogeographic optimization algorithm (KBBO algorithm) are set as follows: the maximum number of iterations Maxgen = 200; the maximum number of populations Popsize = 120; the population size Numvar = 6; the population migration probability Pmodify = 1; the mutation probability Pmutate = 0.01.

[0064] S140. Based on the improved K-means algorithm, the network structure and node parameters of the extreme learning machine network are optimized and selected.

[0065] Specifically, in this step, in the single-output ELM, since the network output weight vector θ is based on the randomly given hidden layer node parameter w i and b i It is calculated that there may be a set of non-optimal or unnecessary parameters and biases of hidden layer nodes. Moreover, compared with general SLFN, ELM tends to select more hidden layer nodes to obtain better learning ability, but for unknown test data, it sometimes affects the performance of the network. Based on the above considerations. The present invention applies the KBBO algorithm to optimize the network structure and node parameters of ELM to further improve the learning performance of ELM, which is called the KBBO-ELM method. The network structure optimization of the KBBO-ELM model includes the selection of network input variables, the number of network hidden layer nodes and the activation function. The node parameter optimization includes the network hidden layer node parameters w i and b i Optimization.

[0066] S150. Set relevant parameters so that the network conforms to the boiler combustion model.

[0067] S160: Based on the clustered real-time operation data, a preset extreme learning machine network is trained to obtain the boiler combustion model.

[0068] S170: Predict boiler combustion thermal efficiency using the boiler combustion model.

[0069] Specifically, in this step, the boiler combustion model is used to predict the actual boiler combustion thermal efficiency, and the training error convergence curve is as follows: Figure 3 shown.

[0070] The boiler combustion thermal efficiency prediction method of this embodiment combines the biogeographic optimization algorithm with the K-means clustering algorithm, aiming to search for the initial cluster center, so that the clustering algorithm has good classification accuracy and global search ability. On the basis of the ELM prediction model, the KBBO optimization algorithm is used for the optimization selection of the network input variables, the configuration and parameters of the hidden layer nodes, and the Tikhonov regularization parameters of the ELM. This method can effectively improve the prediction accuracy and show its effectiveness. Therefore, the boiler combustion thermal efficiency prediction method of this embodiment can effectively improve the prediction accuracy and prediction efficiency.

[0071] Another aspect of the present disclosure is Figure 4 As shown, a boiler combustion thermal efficiency prediction device 100 is provided. The device 100 can be applied to the method described above. The device 100 includes:

[0072] The acquisition module 110 is used to acquire the real-time operation data of the boiler and pre-process the real-time operation data;

[0073] A partitioning module 120, used for partitioning the preprocessed real-time operation data into a training set and a test set;

[0074] The clustering module 130 is used to cluster the real-time operation data in the training set using an improved K-means algorithm, specifically comprising: firstly optimizing the K-means initial clustering center using a biogeographic optimization algorithm, and then clustering the real-time operation data using the K-means algorithm;

[0075] A setting module 140, used to set relevant parameters so that the network conforms to the boiler combustion model;

[0076] A training module 150 is used to train a preset extreme learning machine network based on the clustered real-time operation data to obtain the boiler combustion model;

[0077] The prediction module 160 is used to predict the boiler combustion thermal efficiency using the boiler combustion model.

[0078] The boiler combustion thermal efficiency prediction device of this embodiment combines the biogeographic optimization algorithm with the K-means clustering algorithm, aiming to search for the initial cluster center, so that the clustering algorithm has good classification accuracy and global search ability. On the basis of the ELM prediction model, the KBBO optimization algorithm is used for the optimization selection of the network input variables, the configuration and parameters of the hidden layer nodes, and the Tikhonov regularization parameters of the ELM. This method can effectively improve the prediction accuracy and show its effectiveness. Therefore, the boiler combustion thermal efficiency prediction device of this embodiment can effectively improve the prediction accuracy and prediction efficiency.

[0079] In some implementations, the clustering module 130 is further configured to:

[0080] Initialize the parameters related to the biogeography optimization algorithm;

[0081] Calculate the fitness value of each solution, the number of species, immigration rate and emigration rate, and save individuals with high fitness.

[0082] Use the immigration rate to select the individual variables that need to be migrated in, and use the out-migration rate to select the individual variables that need to be migrated out, and perform the migration operation;

[0083] Perform mutation operations based on mutation probability;

[0084] Whether the iteration stop condition is met, if not, continue to execute the above steps;

[0085] The individuals with high fitness are clustered as the initial center points.

[0086] Another aspect of the present disclosure provides an electronic device, including:

[0087] one or more processors;

[0088] A storage unit is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors can implement the method described above.

[0089] Another aspect of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program can implement the method described above when executed by a processor.

[0090] The computer-readable medium may be included in the apparatus, device, or system of the present disclosure, or may exist independently.

[0091] Among them, computer-readable storage media can be any tangible media that contains or stores a program, which can be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment. More specific examples include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, an optical fiber, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0092] The computer-readable storage medium may also include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code, specific examples of which include but are not limited to electromagnetic signals, optical signals, or any suitable combination thereof.

[0093] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present disclosure, but the present disclosure is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and substance of the present disclosure, and these modifications and improvements are also considered to be within the scope of protection of the present disclosure.

Claims

1. A method for predicting boiler combustion thermal efficiency, characterized in that: The method comprises: Collecting real-time operation data of the boiler and preprocessing the real-time operation data; Divide the preprocessed real-time running data into a training set and a test set; The improved K-means algorithm is used to cluster the real-time operation data in the training set, specifically comprising: firstly optimizing the initial clustering center of K-means by using a biogeographic optimization algorithm, and then clustering the real-time operation data by using the K-means algorithm; Set relevant parameters to make the network conform to the boiler combustion model; Based on the clustered real-time operation data, a preset extreme learning machine network is trained to obtain the boiler combustion model; Predicting boiler combustion thermal efficiency using the boiler combustion model; Before training the extreme learning machine network, the method further includes: Based on the improved K-means algorithm, the network structure and node parameters of the extreme learning machine network are optimized and selected; wherein the network structure optimization includes the selection of network input variables, the number of network hidden layer nodes and activation functions; the node parameter optimization includes the selection of network hidden layer node parameters w i and b i Optimization.

2. The method according to claim 1, characterized in that The biogeography optimization algorithm is used to optimize the initial K-means clustering center, including: Initialize the parameters related to the biogeography optimization algorithm; Calculate each solution and its fitness value, and calculate the corresponding number of species and immigration rate and emigration rate, while saving individuals with high fitness; Use the immigration rate to select the individual variables that need to be migrated in, and use the out-migration rate to select the individual variables that need to be migrated out, and perform the migration operation; Perform mutation operations based on mutation probability; Whether the iteration stop condition is met, if not, continue to execute the above steps; The individuals with high fitness are clustered as the initial center points.

3. The method according to claim 2, characterized in that The parameters in the biogeographic optimization algorithm are set as follows: The maximum number of iterations Maxgen=200; the maximum population size Popsize=120; the population size Numvar=6; the population migration probability Pmodify=1; the mutation probability Pmutate=0.

01.

4. The method according to any one of claims 1 to 3, characterized in that: The preprocessing of the real-time operation data comprises: The real-time operation data is processed in advance by unweighted group average method UPGMA to remove noise data and edge data.

5. A boiler combustion thermal efficiency prediction device, characterized in that: The device comprises: A collection module, used for collecting real-time operation data of the boiler and preprocessing the real-time operation data; A partitioning module is used to partition the preprocessed real-time running data into a training set and a test set; A clustering module is used to cluster the real-time operation data in the training set using an improved K-means algorithm, specifically comprising: firstly optimizing the initial K-means clustering center using a biogeographic optimization algorithm, and then clustering the real-time operation data using the K-means algorithm; A setting module, used to set relevant parameters so that the network conforms to the boiler combustion model; A training module, used for training a preset extreme learning machine network based on the clustered real-time operation data to obtain the boiler combustion model; A prediction module, used to predict the boiler combustion thermal efficiency using the boiler combustion model; Before training the extreme learning machine network, the setting module is also used to: Based on the improved K-means algorithm, the network structure and node parameters of the extreme learning machine network are optimized and selected; wherein the network structure optimization includes the selection of network input variables, the number of network hidden layer nodes and activation functions; the node parameter optimization includes the selection of network hidden layer node parameters w i and b i Optimization.

6. The device according to claim 5, characterized in that The clustering module is further specifically used for: Initialize the parameters related to the biogeography optimization algorithm; Calculate each solution and its fitness value, and calculate the corresponding number of species and immigration rate and emigration rate, while saving individuals with high fitness; Use the immigration rate to select the individual variables that need to be migrated in, and use the out-migration rate to select the individual variables that need to be migrated out, and perform the migration operation; Perform mutation operations based on mutation probability; Whether the iteration stop condition is met, if not, continue to execute the above steps; The individuals with high fitness are clustered as the initial center points.

7. An electronic device, characterized in that: include: one or more processors; A storage unit, used to store one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement the method according to any one of claims 1 to 4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 4 can be implemented.

Images