A method, device and computing equipment for predicting oxygen content of boiler flue gas

Abstract

The present disclosure relates to the technical field of boiler flue gas oxygen content prediction, and provides a boiler flue gas oxygen content prediction method, device and computing equipment. The method comprises: obtaining a parameter-optimized task model of a target boiler, and sample data used for training the parameter-optimized task model, wherein the sample data comprises first sample data of a source domain boiler and sample weight data of the source domain boiler about the target boiler; training the parameter-optimized task model by using the first sample data and the sample weight data to obtain a target prediction model for predicting flue gas oxygen content of the target boiler; and predicting operation parameter data of the target boiler based on the target prediction model to obtain a flue gas oxygen content value of the target boiler. The present disclosure further improves the prediction accuracy of the target prediction model about the flue gas oxygen content value of the target boiler from the improvements of the data sample and the training model.

Description

Technical Field

[0001] This disclosure relates to the field of boiler flue gas oxygen content prediction technology, and in particular to a method, apparatus and computing device for predicting boiler flue gas oxygen content. Background Technology

[0002] In existing technologies, historical boiler operating data or measured operating parameter data are used to learn a predictive model for boiler flue gas oxygen content. This model is then used to automatically predict the optimal flue gas oxygen content under boiler operating conditions, enabling optimized energy consumption control. However, practical applications have revealed that this data-driven model learning method suffers from inaccuracies due to variations in data distribution across different boilers. A model learned from data from one boiler may not accurately predict flue gas oxygen content for another. This inaccuracy stems from both the data source and the model training method. Therefore, improving the accuracy of boiler flue gas oxygen content prediction models remains a key research area. Summary of the Invention

[0003] In view of this, the present disclosure provides a method, apparatus and computing device for predicting the oxygen content of boiler flue gas, in order to solve the problem that the prediction model for learning the oxygen content of boiler flue gas using data from one boiler has low prediction accuracy under another boiler.

[0004] A first aspect of this disclosure provides a method for predicting the oxygen content of boiler flue gas, comprising: acquiring a task model for parameter optimization of a target boiler, and sample data for training the task model for parameter optimization, the sample data including first sample data of source boilers and sample weight data of source boilers with respect to the target boiler; training the task model for parameter optimization using the first sample data and the sample weight data to obtain a target prediction model for predicting the oxygen content of the flue gas of the target boiler; and predicting the operating parameter data of the target boiler based on the target prediction model to obtain the oxygen content value of the flue gas of the target boiler.

[0005] A second aspect of this disclosure provides a boiler flue gas oxygen content prediction device based on model parameter optimization, comprising: an acquisition module configured to acquire a parameter optimization task model of a target boiler, and sample data for training the parameter optimization task model, the sample data including first sample data of source domain boilers and sample weight data of source domain boilers with respect to the target boiler; a training module configured to train the parameter optimization task model using the first sample data and the sample weight data to obtain a target prediction model for predicting the flue gas oxygen content of the target boiler; and a prediction module configured to predict the operating parameter data of the target boiler based on the target prediction model to obtain the flue gas oxygen content value of the target boiler.

[0006] A third aspect of this disclosure provides a computing device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0007] The beneficial effects of this disclosed embodiment compared with the prior art are as follows: by using the first sample data of the source boiler to perform sample transfer to obtain sample data of the task model of the target boiler, the difference in data distribution between the target boiler and the source boiler is reduced, and the task model for the target boiler is a model after parameter optimization. The improvement of both data samples and training model further improves the prediction accuracy of the target prediction model for the oxygen content of flue gas of the target boiler. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic diagram illustrating an application scenario of an embodiment of this disclosure; Figure 2 This is a schematic flowchart of a method for predicting the oxygen content of boiler flue gas provided in an embodiment of this disclosure; Figure 3 This is a flowchart illustrating another method for predicting the oxygen content in boiler flue gas provided in this embodiment of the present disclosure. Figure 4 This is a schematic diagram of the structure of a boiler flue gas oxygen content prediction device provided in an embodiment of this disclosure; Figure 5 This is a schematic diagram of the structure of a computing device provided in an embodiment of this disclosure. Detailed Implementation

[0010] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of this disclosure. However, those skilled in the art will understand that this disclosure may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this disclosure with unnecessary detail.

[0011] A method and apparatus for predicting the oxygen content of boiler flue gas according to an embodiment of the present disclosure will now be described in detail with reference to the accompanying drawings.

[0012] Figure 1 This is a schematic diagram illustrating an application scenario of an embodiment of this disclosure. The application scenario may include a source boiler 1, a target boiler 2, electronic equipment 3, and a network 4.

[0013] Source boiler 1 may include one or more boilers. These source boilers 1 have a relatively accurate flue gas oxygen content prediction model. This flue gas oxygen content prediction model is a machine model trained on a machine learning algorithm using the operating parameter data of source boiler 1 as sample data. It can intelligently calculate the optimal flue gas oxygen content value to control the boiler operation and maintain optimal boiler efficiency. Target boiler 2 is a different boiler from source boiler 1. In practical applications, if the flue gas oxygen content prediction model of source boiler 1 is directly used to predict the target boiler 2, the predicted flue gas oxygen content value may have a large error compared to the actual value of the flue gas oxygen content of target boiler 2, i.e., the prediction accuracy is insufficient.

[0014] Electronic device 3 can communicate with source boiler 1 and target boiler 2 via network 4. Using this communication connection, electronic device 3 can obtain the operating data of source boiler 1 and target boiler 2, analyze and process this operating data, and send the analyzed and processed results to source boiler 1 and target boiler 2.

[0015] Specifically, the electronic device 3 can be either hardware or software. When the electronic device 3 is hardware, it can be various devices that support communication with the source boiler 1 and the target boiler 2, including but not limited to industrial control computers, tablet computers, desktop computers, and servers; when the electronic device 3 is software, it can be installed on the aforementioned hardware devices. The electronic device 3 can be implemented as multiple software programs or software modules, or as a single software program or software module, and this disclosure does not limit this. Furthermore, various applications can be installed on the electronic device 3, such as data processing applications, machine learning models, industrial control platform software, data search applications, etc.

[0016] Network 4 can be a wired network using coaxial cable, twisted pair, and fiber optic connection, or it can be a wireless network that enables interconnection of various communication devices without wiring, such as WiFi and carrier waves. This disclosure does not limit the scope of the embodiments.

[0017] It should be noted that the specific types, quantities and combinations of the source boiler 1, target boiler 2, electronic equipment 3 and network 4 can be adjusted according to the actual needs of the application scenario, and this disclosure embodiment does not limit this.

[0018] Figure 2 This is a flowchart of a method for predicting the oxygen content of boiler flue gas provided in an embodiment of this disclosure. Figure 2The method for predicting the oxygen content in boiler flue gas can be derived from... Figure 1 The electronic device performs this action. For example... Figure 2 As shown, the method for predicting the oxygen content in boiler flue gas includes: S201, Obtain the task model for parameter optimization of the target boiler, and the sample data for training the task model for parameter optimization. The sample data includes the first sample data of the source domain boiler and the sample weight data of the source domain boiler with respect to the target boiler. S202, the task model for parameter optimization is trained using the first sample data and sample weight data to obtain the target prediction model for predicting the oxygen content of the flue gas of the target boiler. S203, based on the target prediction model, predict the operating parameter data of the target boiler to obtain the oxygen content value of the flue gas of the target boiler.

[0019] This embodiment utilizes first sample data from the source boiler to perform sample migration to obtain sample data for the task model of the target boiler, thereby reducing the data distribution differences between the target boiler and the source boiler. Furthermore, the task model for the target boiler is a model with optimized parameters. Improvements in both data samples and training models further enhance the prediction accuracy of the target prediction model for the oxygen content of the flue gas from the target boiler.

[0020] In some embodiments, obtaining sample data for training a task model for parameter optimization, the sample data including first sample data of the source domain boiler and sample weight data of the source domain boiler with respect to the target boiler, includes: obtaining the first sample data of the source domain boiler and second sample data of the target boiler; mixing the first sample data and the second sample data, and using the mixed data to train a kernel density estimation algorithm to obtain a corresponding kernel density estimation model; using the first sample data as data for the kernel density estimation model, and obtaining the sample weight data of the source domain boiler with respect to the target boiler from the output of the kernel density estimation model; and using the sample weight data and the first sample data as sample data for training the task model for parameter optimization.

[0021] Specifically, the first sample data of the source boiler and the second sample data of the target boiler are mixed and used to train the kernel density estimation algorithm. The first sample data of the source boiler is input into the trained model to obtain the sample weight data of the source boiler with respect to the target boiler. This reduces the distribution difference between the data of the source boiler and the target boiler, realizes the sample transfer between the source boiler and the target boiler, improves the sample data quality of the training task model, and thus improves the accuracy of the prediction model of the target boiler.

[0022] Furthermore, the feature data portion of the first and second sample data may include steam boiler flue gas temperature, economizer outlet temperature, instantaneous flue gas flow rate, steam boiler fuel gas temperature, steam boiler standard flue gas flow rate, steam boiler natural gas inlet pressure, steam boiler flue gas velocity, steam boiler condenser inlet flue gas temperature, steam boiler exhaust gas temperature, steam boiler flue gas pressure, steam boiler condenser inlet pressure, steam boiler main steam instantaneous flow rate, steam boiler operating status, and steam boiler natural gas inlet instantaneous flow rate, etc. The target value of the first and second sample data includes the oxygen content of the steam boiler flue gas. That is, the first and second sample datasets can be labeled sample data.

[0023] In some embodiments, obtaining a parameter optimization task model for the target boiler includes: obtaining a task model for the target boiler; determining the optimal parameters of the task model based on an adaptive genetic algorithm, and placing the optimal parameters into the task model to obtain a parameter-optimized task model.

[0024] Specifically, the task model for the target boiler can be a regression-based algorithm or model, that is, using historical operating data of the source boiler and the target boiler as sample data for learning to obtain a predictive model for the oxygen content in the flue gas of the target boiler. In this embodiment, an adaptive genetic algorithm is used to optimize the model parameters of the task model, thereby further improving the accuracy of the model prediction.

[0025] For example, the task model can be a neural network model or an XGBoost model. When the task model is a neural network model, the optimal threshold and weights can be obtained through the operation of an adaptive genetic algorithm. Then, these optimal thresholds and weights are placed into the neural network model to obtain a parameter-optimized task model. When the task model is an XGBoost model, the optimal initial parameters of the model can be obtained through the operation of an adaptive genetic algorithm. Then, these optimal initial parameters are placed into the XGBoost model to obtain a parameter-optimized task model. It should be understood that the specific type of task model is not limited in the embodiments of this disclosure.

[0026] In some embodiments, determining the optimal parameters of the task model based on an adaptive genetic algorithm includes: generating an initial population and a fitness function based on the task model of the target boiler; calculating the fitness value of each individual in the population using the fitness function; sorting all individuals in the population in ascending order of fitness value to obtain a sorted population; dividing the sorted population into three subpopulations (good, medium, and poor) based on a segmented selection strategy, and randomly selecting some individuals from each of the three subpopulations to form a selected population; performing a crossover operation on the individuals in the selected population based on an adaptive crossover probability to generate new individuals; performing a mutation operation on the selected new individuals based on an adaptive mutation probability to generate new individuals, and obtaining a new population from the new individuals; and obtaining the optimal parameters of the task model when the new population meets preset conditions, including the new population reaching a preset generation threshold or the new population's minimum fitness value reaching a preset error precision.

[0027] Specifically, the fitness function of the adaptive genetic algorithm can be: ,in, Represents the actual value. Indicates the predicted value. Indicates the number of samples. This indicates the first in the specific sample data. One sample, It is a positive integer, and ∈[1, ].

[0028] In this embodiment, the adaptive genetic algorithm employs a combination of segmented selection and random sampling in population selection. The population is arranged in ascending order of fitness values, dividing it into three subpopulations: good, average, and poor. Individuals are then selected proportionally from these subpopulations to form the final population. This approach aims to preserve good individuals while ensuring diversity in the next generation. In some embodiments, the preferred ratio for selecting individuals from the good, average, and poor subpopulations is 5:3:2. However, other ratios can be used, and this embodiment does not impose any limitations on this.

[0029] Furthermore, the adaptive genetic algorithm provided in this embodiment of the present disclosure is also adaptive in determining the crossover probability and mutation probability during crossover and mutation operations, thereby effectively ensuring the evolution speed of the genetic algorithm and the quality of the evolving population.

[0030] In some embodiments, crossover operations are performed on individuals in the selected population based on adaptive crossover probabilities to generate new individuals, including: determining a first crossover probability and a second crossover probability based on the minimum fitness value, average fitness value, and the smaller fitness value of the two crossover individuals and an initial crossover probability, wherein the first crossover probability is greater than the second crossover probability; comparing the smaller fitness value of the two crossover individuals with the average fitness value of the current population; if the smaller fitness value is greater than or equal to the average fitness value, performing a crossover operation on the two individuals using the first crossover probability to generate a new individual; if the smaller fitness value is less than the average fitness value, performing a crossover operation on the two individuals using the second crossover probability to generate a new individual.

[0031] Specifically, adaptive crossover operations are performed with a certain probability. The crossover probability is used to recombine individuals and generate new individuals. This adjusted value ensures that desirable genes are preserved as much as possible during the evolution of the offspring population. For example, the crossover probability can be determined using the following formula 1): ...1); in, , These represent the larger and smaller values ​​in the initial crossover probability, respectively. For example, the initial crossover probability is [ , ];in addition, , , These represent the minimum fitness value, average fitness value, and smaller fitness value among the crossover individuals in the current population, respectively.

[0032] The crossover mechanism for two individuals in the population after the selection operation is as follows: ; in: , Individuals selected before the crossover operation. , The new individual generated after the crossover operation. It is a random number in the range [0,1].

[0033] In some embodiments, performing a mutation operation on a selected new individual based on an adaptive mutation probability to generate a new individual includes: determining a first mutation probability and a second mutation probability based on the minimum fitness value, average fitness value, and the fitness value and initial mutation probability of the mutated individual in the current population, wherein the first mutation probability is greater than the second mutation probability; comparing the fitness value of the mutated individual with the average fitness value in the current population; if the fitness value of the mutated individual is greater than or equal to the average fitness value, performing a mutation operation on the individual using the first mutation probability to generate a new individual; if the fitness value of the mutated individual is less than the average fitness value, performing a mutation operation on the mutated individual using the second mutation probability to generate a new individual.

[0034] Specifically, the adaptive mutation operation is performed through a certain probability. (i.e., crossover probability) is used to mutate selected individuals, preserving population diversity, while also using adaptive probability. This is to prevent genetic mutations in high-quality individuals. For example, the mutation probability can be determined using the following formula 2): ...2); in: , These represent the larger and smaller values ​​of the initial mutation probabilities, respectively. For example, the initial crossover probability is [ , ]; , , These represent the minimum fitness, average fitness, and fitness of the mutated individual in the current population, respectively. In the mutation operation, individuals are selected using random sampling. Genes that need to be mutated Then, the following non-uniform mutation operator is used to mutate the gene: in: For genes The gene obtained after mutation, , These represent the current generation and the total number of iterations in the genetic algorithm, respectively. A random number in (0,1) , These represent the minimum and maximum values ​​within the range of values ​​for this variant gene. The threshold at which a population tends to stabilize.

[0035] Furthermore, to avoid the adaptive genetic algorithm from easily converging to local optima, the following simulated annealing re-optimization operation can be performed when the population tends to stabilize.

[0036] In some embodiments, after obtaining a new population, the flue gas oxygen content prediction method further includes: determining the stability of the new population based on the fitness value of the best individual in the new population and the average fitness value of the new population; and optimizing the individuals in the new population using a simulated annealing algorithm when the stability meets a preset condition, to obtain an optimized new population, wherein the preset condition includes that the difference between the fitness value of the best individual in the new population and the average fitness value of the new population is less than a preset value.

[0037] Specifically, the conditions for determining whether a population is stabilizing are as follows: ; in, This represents the average fitness value in the population. This represents the fitness value corresponding to the best individual in the population. When the average fitness value is close to the optimal fitness value, the population evolution tends to be stable. Specifically, based on the fitness function, it can be seen that the smaller the fitness value, the better the individual's genes; conversely, the larger the fitness value, the worse the individual's genes. Therefore, since the embodiments of this disclosure perform simulated annealing operations on the population obtained after crossover and mutation, while crossover and mutation are performed on the selected population, the average fitness value and the minimum fitness value in the two populations may be the same or different.

[0038] Specifically, the simulated annealing re-optimization operation can be performed as follows: First, generate new individuals within the neighborhood of the original individual, assuming the fitness value of the original individual is... The fitness value of the new individual is The probability of a new individual replacing the original individual As shown below: ; in: For the original individual, As a new individual, This is the temperature during the current simulated annealing.

[0039] In this embodiment of the disclosure, after the above-mentioned crossover and mutation operations, the population is continuously evolved to obtain a new population. After each new population is obtained, it is determined whether the current evolutionary generation has reached the maximum evolutionary generation or the minimum fitness value has reached the set error precision. If any of the above conditions are met, the optimal parameters obtained after the improved genetic simulated annealing algorithm operation are saved, and then the parameters in the task model are set to the optimal parameters. Finally, the network is optimized through sample data to obtain the final target prediction model.

[0040] In addition, after obtaining a new population, if the population does not meet the above stability requirements, or if the new population obtained after simulated annealing does not reach the maximum number of generations or the minimum fitness value of the new population does not reach the set error precision, then the step of calculating the fitness value of each individual in the population using the fitness function is returned to perform the iterative operation.

[0041] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0042] Figure 3 This is a flowchart of another method for predicting the oxygen content of boiler flue gas provided in this disclosure. Figure 3 As shown, the method for predicting the oxygen content in boiler flue gas includes: S301, Obtain the task model of the target boiler and the sample data for training the task model. The sample data includes the first sample data of the source domain boiler and the sample weight data of the source domain boiler with respect to the target boiler. S302, the genetic algorithm and simulated annealing algorithm are used to optimize the parameters of the task model of the target boiler to obtain the parameter-optimized task model; S303, using the first sample data and sample weight data to train the parameter optimization task model, and obtain the target prediction model for predicting the oxygen content of the flue gas of the target boiler. S304. Based on the target prediction model, the operating parameter data of the target boiler are predicted to obtain the oxygen content value of the flue gas of the target boiler.

[0043] This embodiment utilizes first sample data from the source boiler to perform sample migration to obtain sample data for the task model of the target boiler, thereby reducing the data distribution differences between the target boiler and the source boiler. Furthermore, the task model for the target boiler is a model with optimized parameters. Improvements in both data samples and training models further enhance the prediction accuracy of the target prediction model for the oxygen content of the flue gas from the target boiler.

[0044] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein. For details not disclosed in the apparatus embodiments of this disclosure, please refer to the embodiments of the method disclosed herein.

[0045] Figure 4 This is a schematic diagram of a boiler flue gas oxygen content prediction device provided in an embodiment of this disclosure. Figure 4 As shown, the boiler flue gas oxygen content prediction device includes: The acquisition module 401 is configured to acquire a task model for parameter optimization of the target boiler, and sample data for training the task model for parameter optimization. The sample data includes first sample data of the source domain boiler and sample weight data of the source domain boiler with respect to the target boiler. Training module 402 is configured to train the parameter-optimized task model using the first sample data and sample weight data to obtain a target prediction model for predicting the oxygen content of flue gas in the target boiler. The prediction module 403 is configured to predict the operating parameter data of the target boiler based on the target prediction model to obtain the oxygen content value of the flue gas of the target boiler.

[0046] According to the technical solution provided in this disclosure, sample data for the task model of the target boiler is obtained by sample migration through the first sample data of the source boiler, which reduces the data distribution difference between the target boiler and the source boiler. Furthermore, the task model for the target boiler is a model after parameter optimization. Improvements in both data samples and training models further enhance the prediction accuracy of the target prediction model for the oxygen content of flue gas in the target boiler.

[0047] In some embodiments, Figure 4 The acquisition module 401 acquires the task model of the target boiler; based on the adaptive genetic algorithm, it determines the optimal parameters of the task model and places the optimal parameters into the task model to obtain the parameter-optimized task model.

[0048] In some embodiments, the boiler flue gas oxygen content prediction device further includes: a model optimization module 404, configured to generate an initial population and fitness function based on a task model of the target boiler; calculate the fitness value of each individual in the population using the fitness function; sort all individuals in the population in ascending order of fitness value to obtain a sorted population; divide the sorted population into three sub-populations (good, medium, and poor) based on a segmented selection strategy, and randomly select some individuals from each of the three sub-populations to form a selected population; perform crossover operation on the individuals in the selected population based on an adaptive crossover probability to generate new individuals; perform mutation operation on the selected new individuals based on an adaptive mutation probability to generate new individuals, and obtain a new population from the new individuals; and obtain the optimal parameters of the task model when the new population meets preset conditions, including the new population's evolutionary generation reaching a preset generation threshold or the new population's minimum fitness value reaching a preset error accuracy.

[0049] In some embodiments, Figure 4The model optimization module 404 determines the first crossover probability and the second crossover probability of the crossover operation based on the minimum fitness value, the average fitness value, the smaller fitness value of the two crossover individuals, and the initial crossover probability in the current population. The first crossover probability is greater than the second crossover probability. It compares the smaller fitness value of the two crossover individuals with the average fitness value in the current population. If the smaller fitness value is greater than or equal to the average fitness value, it performs the crossover operation on the two individuals using the first crossover probability to generate a new individual. If the smaller fitness value is less than the average fitness value, it performs the crossover operation on the two individuals using the second crossover probability to generate a new individual.

[0050] In some embodiments, Figure 4 The model optimization module 404 determines a first mutation probability and a second mutation probability for the mutation operation based on the minimum fitness value, average fitness value, and the fitness value and initial mutation probability of the mutated individual in the current population. The first mutation probability is greater than the second mutation probability. The module compares the fitness value of the mutated individual with the average fitness value in the current population. If the fitness value of the mutated individual is greater than or equal to the average fitness value, the module performs the mutation operation on the individual using the first mutation probability to generate a new individual. If the fitness value of the mutated individual is less than the average fitness value, the module performs the mutation operation on the mutated individual using the second mutation probability to generate a new individual.

[0051] In some embodiments, after obtaining a new population Figure 4 The model optimization module 404 determines the stability of the new population based on the fitness value of the best individual in the new population and the average fitness value of the new population. If the stability meets the preset conditions, the simulated annealing algorithm is used to optimize the individuals in the new population to obtain the optimized new population. The preset conditions include that the difference between the fitness value of the best individual in the new population and the average fitness value of the new population is less than a preset value.

[0052] In some embodiments, Figure 4 The acquisition module 401 acquires the first sample data of the source boiler and the second sample data of the target boiler; mixes the first sample data and the second sample data, and uses the mixed data to train a kernel density estimation algorithm to obtain the corresponding kernel density estimation model; uses the first sample data as the data of the kernel density estimation model, and obtains the sample weight data of the source boiler with respect to the target boiler from the output of the kernel density estimation model; and uses the sample weight data and the first sample data as the sample data of the task model for training parameter optimization.

[0053] In some embodiments, the task model includes a neural network model and an XGBoost model.

[0054] In some embodiments, the boiler flue gas oxygen content prediction device provided in this disclosure may further include: an acquisition module 401, configured to acquire a task model of a target boiler and sample data for training the task model, the sample data including first sample data of source boilers and sample weight data of source boilers with respect to the target boiler; a model optimization module 404, configured to optimize the parameters of the task model of the target boiler using a genetic algorithm and a simulated annealing algorithm to obtain a parameter-optimized task model; a training module 402, configured to train the parameter-optimized task model using the first sample data and sample weight data to obtain a target prediction model for predicting the flue gas oxygen content of the target boiler; and a prediction module 403, configured to predict the operating parameter data of the target boiler based on the target prediction model to obtain the flue gas oxygen content value of the target boiler.

[0055] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this disclosure.

[0056] Figure 5 This is a schematic diagram of the computing device 5 provided in an embodiment of this disclosure. Figure 5 The computing device 5 in the middle can be applied to Figure 1 In electronic devices, such as Figure 5 As shown, the computing device 5 of this embodiment includes a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, it implements the steps in the various method embodiments described above. Alternatively, when the processor 501 executes the computer program 503, it implements the functions of each module / unit in the various device embodiments described above.

[0057] Exemplarily, computer program 503 may be divided into one or more modules / units, which are stored in memory 502 and executed by processor 501 to perform the present disclosure. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 503 in computing device 5.

[0058] The computing device 5 may be an electronic device such as a desktop computer, laptop, handheld computer, or cloud server. The computing device 5 may include, but is not limited to, a processor 501 and a memory 502. Those skilled in the art will understand that... Figure 5This is merely an example of computing device 5 and does not constitute a limitation on computing device 5. It may include more or fewer components than shown, or combine certain components, or different components. For example, computing device may also include input / output devices, network access devices, buses, etc.

[0059] Processor 501 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0060] The memory 502 can be an internal storage unit of the computing device 5, such as a hard disk or RAM of the computing device 5. The memory 502 can also be an external storage device of the computing device 5, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computing device 5. Furthermore, the memory 502 can include both internal storage units and external storage devices of the computing device 5. The memory 502 is used to store computer programs and other programs and data required by the computing device. The memory 502 can also be used to temporarily store data that has been output or will be output.

[0061] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0062] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0063] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0064] In the embodiments provided in this disclosure, it should be understood that the disclosed apparatus / computing device and method can be implemented in other ways. For example, the apparatus / computing device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0065] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0066] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0067] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in a computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0068] The above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit it. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be included within the protection scope of this disclosure.

Claims

1. A method for predicting the oxygen content in boiler flue gas, characterized in that, include: A task model for parameter optimization of the target boiler is obtained, as well as sample data for training the task model for parameter optimization. The sample data includes first sample data of the source domain boilers and sample weight data of the source domain boilers with respect to the target boiler. The task model for parameter optimization is trained using the first sample data and the sample weight data to obtain a target prediction model for predicting the oxygen content of flue gas in the target boiler. Based on the target prediction model, the operating parameter data of the target boiler are predicted to obtain the oxygen content value of the flue gas of the target boiler. The task model for obtaining the parameter optimization of the target boiler includes: Obtain the task model of the target boiler; Based on the task model of the target boiler, an initial population and fitness function are generated; The fitness value of each individual in the population is calculated using the fitness function. All individuals in the population are sorted in ascending order of fitness value to obtain the sorted population. Based on the segmented selection strategy, the sorted population is divided into three subpopulations: good, medium, and poor. Individuals are randomly selected from each of the three subpopulations to form the selected population. Based on an adaptive crossover probability, crossover operations are performed on individuals in the selected population to generate new individuals; Based on the adaptive mutation probability, the selected new individuals are mutated to generate new individuals, and a new population is obtained from the new individuals. Under the condition that the new population meets the preset conditions, the optimal parameters of the task model are obtained. The preset conditions include the number of generations of the new population reaching a preset generation threshold or the minimum fitness value of the new population reaching a preset error accuracy.

2. The method according to claim 1, characterized in that, The step of performing a crossover operation on individuals in the selected population based on an adaptive crossover probability to generate new individuals includes: Based on the minimum fitness value, average fitness value, and smaller fitness value of the two crossover individuals and the initial crossover probability, determine the first crossover probability and the second crossover probability of the crossover operation, wherein the first crossover probability is greater than the second crossover probability. Compare the smaller fitness value of the two crossover individuals with the average fitness value of the current population; If the smaller fitness value is greater than or equal to the average fitness value, a crossover operation is performed on the two individuals using the first crossover probability to generate a new individual. If the smaller fitness value is less than the average fitness value, a crossover operation is performed on the two individuals using the second crossover probability to generate a new individual.

3. The method according to claim 2, characterized in that, The process of performing mutation operations on the selected new individuals based on adaptive mutation probabilities to generate new individuals includes: Based on the minimum fitness, average fitness, fitness of the mutated individual, and initial mutation probability in the current population, a first mutation probability and a second mutation probability are determined, wherein the first mutation probability is greater than the second mutation probability. Compare the fitness value of the mutated individual with the average fitness value of the current population; If the fitness value of the mutated individual is greater than or equal to the average fitness value, a mutation operation is performed on the individual using the first mutation probability to generate a new individual; If the fitness value of the mutated individual is less than the average fitness value, a mutation operation is performed on the mutated individual using the second mutation probability to generate a new individual.

4. The method according to claim 2, characterized in that, After obtaining the new population, the following is also included: The stability of the new population is determined based on the fitness value of the best individual in the new population and the average fitness value of the new population. If the stability meets the preset conditions, the individuals in the new population are optimized using the simulated annealing algorithm to obtain an optimized new population. The preset conditions include that the difference between the fitness value of the best individual in the new population and the average fitness value of the new population is less than a preset value.

5. The method according to claim 1, characterized in that, The step of acquiring sample data for training the task model to optimize the parameters includes first sample data of the source domain boilers and sample weight data of the source domain boilers with respect to the target boiler, including: Acquire the first sample data of the source boiler and the second sample data of the target boiler; The first sample data and the second sample data are mixed, and a kernel density estimation algorithm is trained using the mixed data to obtain the corresponding kernel density estimation model; The first sample data is used as the data for the kernel density estimation model, and the sample weight data of the source domain boiler with respect to the target boiler is obtained from the output of the kernel density estimation model. The sample weight data and the first sample data are used as sample data for training the task model for parameter optimization.

6. The method according to any one of claims 1-5, characterized in that, The task model includes a neural network model and an XGBoost model.

7. A device for predicting the oxygen content of boiler flue gas, characterized in that, include: The acquisition module is configured to acquire a parameter optimization task model for the target boiler, and sample data for training the parameter optimization task model, wherein the sample data includes first sample data of the source domain boilers and sample weight data of the source domain boilers with respect to the target boiler. The training module is configured to train the parameter-optimized task model using the first sample data and the sample weight data to obtain a target prediction model for predicting the oxygen content of the flue gas of the target boiler. The prediction module is configured to predict the operating parameter data of the target boiler based on the target prediction model to obtain the oxygen content value of the flue gas of the target boiler. The acquisition module is specifically configured to: acquire the task model of the target boiler; generate an initial population and a fitness function based on the task model of the target boiler; and calculate the fitness value of each individual in the population using the fitness function. All individuals in the population are sorted in ascending order of fitness value to obtain the sorted population. Based on a segmented selection strategy, the sorted population is divided into three subpopulations: good, medium, and poor. Individuals are randomly selected from each of these subpopulations to form the selected population. An adaptive crossover operation is performed on the selected population to generate new individuals. An adaptive mutation operation is then performed on the selected new individuals to generate new individuals, and a new population is formed from these new individuals. When the new population meets preset conditions, the optimal parameters of the task model are obtained. These preset conditions include the new population reaching a preset generation threshold or the new population's minimum fitness reaching a preset error precision.

8. A computing device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

Images