A method, device and storage medium for predicting thermal power equipment efficiency

Abstract

The present invention relates to the technical field of thermal power equipment, and in particular to a method, device, equipment and computer storage medium for predicting the efficiency of thermal power equipment. The thermal power equipment efficiency prediction method described in the present invention, based on the analysis of the factors affecting the desulfurization efficiency of the WFGD system, adopts principal component analysis to obtain dimensionality reduction data, and adopts a hybrid strategy to improve the Seagull optimization algorithm for realizing the parameter optimization of the improved support vector machine, thereby constructing a WFGD system desulfurization efficiency prediction model based on ISOA-SVM, and the obtained model can be used to predict the desulfurization efficiency of the WFGD system of an actual thermal power plant.

Description

Technical Field

[0001] The present invention relates to the technical field of thermal power equipment, and in particular to a method, device, equipment and computer storage medium for predicting the efficiency of thermal power equipment. Background Art

[0002] At present, with the continuous growth of global energy demand, the installed capacity of thermal power plants has reached a new high. In the process of thermal power generation, the combustion of coal will produce a large amount of SO2 gas, which is one of the main sources of air pollution. In order to reduce the emission of SO2, thermal power plants have adopted a variety of desulfurization technologies. Limestone-gypsum wet flue gas desulfurization (WFGD) is one of the main desulfurization technologies currently used in thermal power plants. This technology sprays limestone slurry into the flue gas, reacts with SO2 in the flue gas, generates gypsum and discharges it. The desulfurization efficiency of the WFGD system directly affects the economic efficiency of thermal power plant power generation. By predicting the desulfurization efficiency, SO2 emissions can be better controlled and the impact on the environment can be reduced. At the same time, it can also provide a reference for the economic operation of thermal power plants. In order to accurately predict the desulfurization efficiency of the WFGD system, advanced monitoring technology and model algorithms are required.

[0003] However, there are many factors in the WFGD system that can affect the desulfurization efficiency of the system, and these factors often have highly nonlinear and coupled correlations, which makes it difficult to accurately predict the efficiency of the WFGD system. A large number of studies have been carried out in academia and industry on the prediction of desulfurization efficiency and certain results have been achieved. In view of the shortcomings of the prediction performance of BP neural network technology, the prior art has designed an adaptive differential evolution algorithm to optimize the parameters of the BP neural network, thereby obtaining the optimal desulfurization efficiency prediction network model. The prediction accuracy and computational efficiency of this model have been improved compared with the BP neural network. The prior art has established a WFGD desulfurization efficiency prediction model based on partial least squares support vector machine, and selected actual monitoring data to complete model training and testing, and achieved good results. The prior art uses data-driven extreme learning machines to construct a desulfurization system efficiency prediction model, and uses actual power plant measured data for testing. The results show that the model can complete the reasonable prediction of the desulfurization system efficiency, which has a certain reference value for thermal power plant operators. The existing technology builds a regression model of desulfurization system efficiency based on XGBoost, and designs MPGA to predict the desulfurization efficiency when the desulfurization variable cost reaches the minimum value. The test results show that the XGBoost-MPGA regression model has good prediction performance and can take into account stability and convergence. However, the prediction accuracy of the above methods is still difficult to meet the needs of actual engineering, so new methods are urgently needed to further improve the prediction accuracy of desulfurization efficiency. Summary of the invention

[0004] Therefore, the technical problem to be solved by the present invention is to overcome the problem of poor prediction accuracy in the prior art.

[0005] In order to solve the above technical problems, the present invention provides a method for predicting the efficiency of thermal power equipment, comprising:

[0006] Step 1: Obtain data on factors affecting desulfurization efficiency, establish training sets and test sets, and perform data dimensionality reduction;

[0007] Step 2: Initialize the SVM model and use chaos to initialize ISOA population parameters to generate the initialization population;

[0008] Step 3: Input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold and solve the fitness value;

[0009] Step 4: Update the individual positions of seagulls based on combined attack and adversarial learning and re-solve the fitness value;

[0010] Step 5: When the algorithm iteration limit is reached or the optimal fitness value is obtained, the penalty factor and the optimal parameter set of the kernel function are output to the SVM model to obtain a trained SVM model, the test set is input into the trained SVM model, and the prediction result is output.

[0011] Preferably, when the algorithm iteration upper limit is not reached or the optimal fitness value is not obtained, return to step 4.

[0012] Preferably, the data dimension reduction process includes:

[0013] Performing sample standardization processing on the desulfurization efficiency influencing factor data to obtain a correlation coefficient matrix, eigenvalues ​​and eigenvectors of the samples;

[0014] Calculating the contribution rate of each component of the desulfurization efficiency influencing factor data;

[0015] The eigenvector of the corresponding principal component whose contribution rate is greater than the preset threshold is multiplied by the original data to obtain the characteristic matrix of the factors affecting the desulfurization efficiency after dimensionality reduction.

[0016] Preferably, the method of using chaos to initialize ISOA population parameters to generate an initialized population includes:

[0017] A standard SOA population is randomly generated, and the distribution of the standard SOA population is optimized by using a Sinusoidal chaotic mapping to obtain the initialization population.

[0018] Preferably, the updating of individual seagull positions based on combined attack and adversarial learning comprises:

[0019] The sine-cosine algorithm is introduced to improve the attack behavior of individual seagulls. After the individual seagull arrives at a new position, the opposite solution of the inertia weight is introduced for each individual seagull to update the individual position of the seagull.

[0020] Preferably, the formula for improving the attack behavior of individual seagulls by introducing the sine-cosine algorithm is expressed as:

[0021]

[0022] X attack (t) = X new (t)·x·y·z+X best (t)

[0023]

[0024] Among them, X(t) represents the current location of the seagull individual, X attack (t) represents the attack point of the seagull, rand∈[0,1] represents a random number, X best (t) represents the current optimal position of the seagull individual, X new (t) represents the new position that the individual arrives at, r represents the spiral radius of each spiral motion of the seagull individual, ρ∈[0,2π] represents the random angle, and u and v represent the relevant constants of the spiral motion.

[0025] Preferably, after the individual seagull arrives at the new position, an opposing solution of the inertia weight is introduced for each individual seagull, and the calculation formula for updating the individual seagull position is:

[0026]

[0027] Among them, X renew (t) is the updated position of the individual seagull, UB and LB represent the upper and lower limits of the optimization parameters, T max represents the upper limit of iteration, and t represents the current iteration number.

[0028] The present invention also provides a thermal power equipment efficiency prediction device, comprising:

[0029] The data acquisition module is used to obtain the data of factors affecting desulfurization efficiency, establish training sets and test sets, and perform data dimension reduction processing;

[0030] Model initialization module, used to initialize the SVM model and use chaos to initialize ISOA population parameters to generate the initialization population;

[0031] A fitness value calculation module is used to input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold, and solve the fitness value;

[0032] The fitness value update module is used to update the individual positions of seagulls and re-solve the fitness value based on the combined attack and adversarial learning;

[0033] The prediction module is used to output the optimal parameter set of the penalty factor and the kernel function to the SVM model when the algorithm iteration upper limit is reached or the optimal fitness value is obtained, so as to obtain a trained SVM model, input the test set into the trained SVM model, and output the prediction result.

[0034] The present invention also provides a thermal power equipment efficiency prediction device, comprising:

[0035] Memory for storing computer programs;

[0036] A processor is used to implement the above-mentioned steps of a method for predicting the efficiency of thermal power equipment when executing the computer program.

[0037] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for predicting the efficiency of thermal power equipment are implemented.

[0038] The above technical solution of the present invention has the following advantages compared with the prior art:

[0039] The method for predicting the efficiency of thermal power equipment described in the present invention, based on the analysis of the factors affecting the desulfurization efficiency of the WFGD system, adopts principal component analysis to obtain dimensionality reduction data, and adopts a hybrid strategy to improve the Seagull optimization algorithm for realizing the parameter optimization of the improved support vector machine, thereby constructing a WFGD system desulfurization efficiency prediction model based on ISOA-SVM. The obtained model can be used to predict the desulfurization efficiency of the WFGD system in an actual thermal power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0041] Figure 1 It is a flow chart of the implementation of a method for predicting the efficiency of thermal power equipment provided by the present invention;

[0042] Figure 2 It is the WFGD system operation flow chart;

[0043] Figure 3 It is the computational flow chart of ISOA-SVM;

[0044] Figure 4 This is a diagram showing the prediction results. DETAILED DESCRIPTION

[0045] The core of the present invention is to provide a method, device, equipment and computer storage medium for predicting the efficiency of thermal power equipment, which effectively improves the prediction accuracy.

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0047] Please refer to Figure 1 , Figure 1 This is a flow chart of the implementation of a thermal power equipment efficiency prediction method provided by the present invention; the specific operation steps are as follows:

[0048] S101: Obtain data on factors affecting desulfurization efficiency, establish a training set and a test set, and perform data dimension reduction processing;

[0049] S102: Initialize the SVM model and use chaos to initialize ISOA population parameters to generate an initialization population;

[0050] S103: input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold, and solve the fitness value;

[0051] S104: Based on the combined attack and adversarial learning, the individual positions of the seagulls are updated and the fitness values ​​are re-solved;

[0052] S105: When the algorithm iteration limit is reached or the optimal fitness value is obtained, the penalty factor and the optimal parameter set of the kernel function are output to the SVM model to obtain a trained SVM model, the test set is input into the trained SVM model, and the prediction result is output.

[0053] When the algorithm iteration upper limit is not reached or the optimal fitness value is not obtained, return to step S104.

[0054] Based on the above embodiments, this embodiment describes step S101 in detail:

[0055] The operation process of the WFGD system can be described as follows: the original flue gas generated by the boiler of the thermal power plant flows through the booster fan, and is sucked into the absorption tower after the flue gas heat exchange device completes cooling. The flue gas in the tower reacts chemically with the atomized limestone slurry to separate from the SO2 gas, and the slurry droplets are filtered out by the demister. The obtained clean flue gas flows through the flue gas heat exchange device and is heated and discharged into the atmosphere through the chimney. The slurry that previously reacted chemically with the SO2 gas accumulates inside the reaction tank at the bottom of the tower, and at the same time, the slurry generated by the limestone slurry pump and the circulating slurry pump will repeatedly spray from the spray layer to separate from the SO2 gas. At this time, under the action of the oxidation fan, compressed air is transported to the reaction tank to achieve forced oxidation, thereby obtaining the derivative product gypsum. The operation process of the WFGD system is as follows: Figure 2 shown.

[0056] In the WFGD system, the factors that affect the system desulfurization efficiency are generally structural design, input flue gas, absorbent and operating parameters, as shown in Table 1.

[0057] Table 1 Influencing factors of WFGD system

[0058] Tab.1 Influencing factors of WFGD system

[0059]

[0060]

[0061] For the influencing factors of the WFGD system, different influencing factors have different influence weights on the prediction results. Therefore, it is necessary to perform data dimensionality reduction on the influencing factors of the WFGD system. This paper uses principal component analysis (PCA) to perform data dimensionality reduction on the influencing factors of the WFGD system.

[0062] There is a certain correlation between the influencing factors of the WFGD system. Therefore, it is difficult to achieve accurate measurement using traditional methods such as a single chemical reaction model in the efficiency prediction problem. In addition, there is coupling between different factors, so it is necessary to reorganize the correlated factors into multiple unrelated variables.

[0063] The steps of data dimension reduction of factors affecting desulfurization efficiency based on PCA are as follows:

[0064] The data of the factors affecting desulfurization efficiency are subjected to sample standardization to obtain the correlation coefficient matrix, eigenvalue and eigenvector of the samples:

[0065] Standardization of data on factors affecting desulfurization efficiency:

[0066]

[0067] In formula (1): i = 1, 2, ..., p, j = 1, 2, ..., q.

[0068] Calculate the contribution rate of each component of the desulfurization efficiency influencing factor data:

[0069] After the data samples of factors affecting desulfurization efficiency are standardized, the sample correlation coefficient matrix, eigenvalue and eigenvector are formed. The contribution rate C of each component is solved based on the following formula r , determine the final number of variables according to the required information retention:

[0070]

[0071] The eigenvector of the corresponding principal component whose contribution rate is greater than the preset threshold is multiplied by the original data to obtain the characteristic matrix of the factors affecting desulfurization efficiency after dimensionality reduction:

[0072] The data is reduced in dimension according to the result of dot multiplication of the eigenvector of the principal component and the original data. The dimension reduction process is shown in the following formula:

[0073]

[0074] After the dimensionality reduction operation, the data dimension is reduced from P to K. The feature matrix formed after the data dimensionality reduction can be used for sample input.

[0075] Based on the above embodiments, this embodiment describes step S102 in detail:

[0076] SVM is a typical machine learning algorithm. Its main mechanism is to optimize the problem by solving the maximum classification hyperplane in the data sample. SVM generally uses kernel functions to map nonlinear raw data samples to a high-dimensional space and make the samples linearly separable. Therefore, SVM usually has better classification effects on small nonlinear samples.

[0077] In SVM, for the set of data samples {(x1,y1), (x2,y2), …, (x i ,y i )} can be mapped to a high-dimensional space and the optimization problem can be solved. The optimization problem can be expressed as:

[0078]

[0079]

[0080] In formula (4), W represents the weight vector, η represents the deviation value of the fitting; C s represents the penalty parameter; η1, η i represents the relaxation factor; ε represents the loss function; k represents the kernel function; S represents the objective function.

[0081] The present invention designs Lagrange multipliers to improve the optimization problem, which can be expressed as:

[0082]

[0083] In formula (5), ψ represents the dual problem of the original optimization problem.

[0084] In the process of optimizing the data of factors affecting actual desulfurization efficiency, the optimal solution to the problem can be obtained by selecting reasonable kernel functions and penalty parameters.

[0085] In order to avoid the shortcomings of manual parameter screening, this paper adopts a hybrid strategy to improve the Seagull optimization algorithm to optimize the SVM parameters.

[0086] SOA is an intelligent optimization algorithm that imitates the living habits of seagull populations. The implementation process of the algorithm is divided into the following two steps:

[0087] 1) Migration behavior

[0088] Seagull individuals actually perform global search during migration. In order to prevent collisions with other individuals, additional variables are introduced. To solve the new position of the individual. The migration behavior of seagull individuals can be expressed as:

[0089]

[0090]

[0091] In formula (6) and formula (7): X * (t) represents the new position of the seagull individual without position collision, X(t) represents the current position of the seagull individual, t represents the current iteration number, Indicates the migration behavior of setting the search threshold. C f Represents the control parameter, which gradually decays from 2 to 0 during the operation. max Represents the upper limit of iterations.

[0092] In the case of not colliding with other seagulls, the seagull will fly towards the optimal position, which can be expressed as:

[0093]

[0094]

[0095] In formula (8) and formula (9): represents the direction of the optimal position, θ represents the random number used to balance the global search and the local search, R d ∈[0,1] is a random number.

[0096] After the seagulls migrate to a position where no collision will occur, they will fly in the direction of the optimal position and arrive at the new position, which can be expressed as:

[0097]

[0098] In formula (10): X new (t) represents the new position reached by the individual.

[0099] 2) Aggressive behavior

[0100] When foraging, seagulls will attack prey, changing the angle and speed of attacking prey through spiral motion. This behavior can be expressed as:

[0101]

[0102] In formula (11), r represents the spiral radius of each spiral motion of the seagull individual, ρ∈[0,2π] represents the random angle, and u and v represent the related constants of the spiral motion. At this time, the attack position of the seagull individual changes to:

[0103] X attack (t) = X new (t)·x·y·z+X best (t) (12)

[0104] In formula (12): X attack (t) represents the attack point of the individual seagull, X best (t) represents the current optimal position of the individual seagull.

[0105] The standard SOA algorithm is prone to fall into local optimality during operation. This paper uses a hybrid strategy consisting of Sinusoidal chaotic mapping, combined attack and adversarial learning to improve SOA. The improvement process is as follows:

[0106] 1) Sinusoidal Chaotic Mapping

[0107] In standard SOA, the population is randomly generated, which makes the algorithm blind and has low convergence efficiency. Sinusoidal chaotic mapping can effectively optimize the population distribution and strengthen the population diversity in the initial stage, which can be expressed as:

[0108]

[0109] In formula (13), σ is taken as 2.3. k and x k+1 They represent the value of the kth iteration and the new value calculated and solved respectively.

[0110] Based on the above embodiments, this embodiment describes step S104 in detail:

[0111] 2) Combination attack

[0112] The standard SOA uses a single spiral motion optimization, which is easy to fall into the local optimum. This paper introduces the sine-cosine algorithm to improve the individual attack behavior of seagulls:

[0113]

[0114] In formula (14), rand∈[0,1] represents a random number.

[0115] 3) Adversarial Learning

[0116] After the individual positions of the seagulls are updated, the opposite solution of the inertia weight is introduced for each individual seagull and calculated, which can be expressed as:

[0117]

[0118] In formula (15), UB and LB represent the upper and lower limits of the optimization parameters respectively, and X renew (t) Guarantee the location of individual updates.

[0119] The present invention uses ISOA for SVM parameter optimization to obtain the ISOA-SVM model, which is used for efficiency prediction of thermal power desulfurization system. The calculation process is as follows: Figure 3 shown.

[0120] The specific steps of ISOA-SVM are:

[0121] Step 1: Set the corresponding training set and test set based on the data of the desulfurization system affecting the desulfurization efficiency, perform standardization and feature extraction, and input the training set for training.

[0122] Step 2: Initialize SVM, use chaos to initialize ISOA population parameters, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold and solve the fitness value.

[0123] Step 3: Update the individual positions of the seagulls based on the combined attack and adversarial learning and re-solve the fitness value.

[0124] Step 4: Determine whether the algorithm iteration limit or the most suitable value has been reached. If so, output the optimal parameter set. Otherwise, return to step 3.

[0125] Step 5: Substitute the output optimal parameter set into the SVM to obtain the trained SVM.

[0126] Step 6: Use the test set for SVM testing, output the prediction results, and the process ends.

[0127] Based on the above embodiments, the experimental data used in this embodiment is derived from the absorption tower data of a thermal power plant in 2022. The sampling time is from November 7 to November 16, the sampling interval is set to 10 minutes, and a total of 1440 sets of data are obtained.

[0128] The operating system used in the test is Windows 10 64-bit system, the test software used is Matlab2023A, the CPU is Intel Core i9-10900K@5.30GH, the memory is 16GB, and the hard disk capacity is 2TB.

[0129] In this experimental test, the training and testing of the data model affecting desulfurization efficiency were completed based on the cross-validation method, and the training set and test set were divided according to a ratio of 9:1, and a training set containing 1296 groups of data affecting desulfurization efficiency and a test set containing 144 groups of data affecting desulfurization efficiency were obtained. Since there are numerical differences in the indicators in Table 1, they need to be preprocessed first, and then the data dimension reduction is performed based on PCA, and the parameters obtained are shown in Table 2.

[0130] Table 2 Principal component eigenvalues ​​and contribution rates

[0131] Tab.2 Principal Component Eigenvalues ​​and Contribution Rates

[0132]

[0133] It can be seen from Table 2 that the first 12 principal components have more than 90% of the information content of the original 21 indicators, which can effectively characterize the original data characteristics that affect the desulfurization efficiency. Therefore, the characteristics composed of the first 12 influencing factors are extracted to serve as the input of the desulfurization efficiency prediction model.

[0134] Based on formula (3), the dimension reduction of data affecting desulfurization efficiency was completed, and the training set containing 1296 groups of data was input into ISOA-SVM for training after PCA preprocessing.

[0135] In order to verify the effectiveness of the ISOA-SVM model proposed in this paper, the ISOA-SVM model proposed in this paper is compared with BPNN, SVM, GA-SVM, and SOA-SVM. The test set is input into the trained model, and the results are as follows: Figure 4 shown.

[0136] Depend on Figure 4It can be seen that the ISOA-SVM model proposed in this paper predicts the desulfurization efficiency of the desulfurization system closer to the actual efficiency. In order to more scientifically compare the advantages and disadvantages of the models, this embodiment introduces the mean absolute error (MAPE) and the root mean square error (RMSE) to describe the model performance, and the calculation results are shown in Table 3.

[0137] Table 3 Comparison of model prediction results

[0138] Tab.3 Comparison Table of Model Prediction Results

[0139]

[0140] According to Table 3, the BPNN network results are simple, and the learning effect for nonlinear small sample data is poor. Therefore, there is a certain gap between the prediction effect of the desulfurization efficiency of the desulfurization system and SVM. Although SVM is suitable for the processing of nonlinear small sample data, due to the defects of manual parameter selection, its own prediction effect is difficult to meet the needs of practical applications, and its prediction error is relatively high. In view of the shortcomings of manual parameter selection, the GA algorithm can improve this problem to a certain extent, but due to the complexity of the GA algorithm itself, it often causes SVM to fall into local optimality, making the prediction accuracy of the desulfurization efficiency of the desulfurization system lower than that of the SVM optimized by the SOA algorithm with a simple principle. The ISOA algorithm proposed in this paper overcomes the shortcomings of the original SOA algorithm, further improves the diversity of the population, avoids the local optimal problem as much as possible, and better solves the parameter optimization problem of SVM. The final ISOA-SVM prediction accuracy is higher than other methods, and the overall prediction effect of the desulfurization efficiency of the desulfurization system is better.

[0141] Aiming at the problem of insufficient prediction accuracy of desulfurization efficiency of current WFGD system, the present invention proposes a desulfurization efficiency prediction model based on ISOA-SVM. Based on the analysis of the factors affecting the desulfurization efficiency of WFGD system, principal component analysis is used to obtain dimensionality reduction data, and a hybrid strategy is used to improve the Seagull optimization algorithm for parameter optimization of improved support vector machine, thereby constructing a WFGD system desulfurization efficiency prediction model based on ISOA-SVM. The obtained model can be used to predict the desulfurization efficiency of WFGD system in actual thermal power plants.

[0142] The experimental test results show that the average absolute percentage error MAPE of the obtained model in the experimental test is only 0.2783, and the root mean square error RMSE is only 0.2899. The prediction result is very close to the actual value, the prediction accuracy is high, and a good prediction effect is achieved. The desulfurization efficiency prediction method proposed in the present invention can effectively predict the desulfurization efficiency of the desulfurization system of a thermal power plant, has good engineering practical application value, and is helpful to the operation and maintenance work of thermal power plant operators.

[0143] The embodiment of the present invention further provides a thermal power equipment efficiency prediction device; the specific device may include:

[0144] The data acquisition module is used to obtain the data of factors affecting desulfurization efficiency, establish training sets and test sets, and perform data dimension reduction processing;

[0145] Model initialization module, used to initialize the SVM model and use chaos to initialize ISOA population parameters to generate the initialization population;

[0146] A fitness value calculation module is used to input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold, and solve the fitness value;

[0147] The fitness value update module is used to update the individual positions of seagulls and re-solve the fitness value based on the combined attack and adversarial learning;

[0148] The prediction module is used to output the optimal parameter set of the penalty factor and the kernel function to the SVM model when the algorithm iteration upper limit is reached or the optimal fitness value is obtained, so as to obtain a trained SVM model, input the test set into the trained SVM model, and output the prediction result.

[0149] The thermal power equipment efficiency prediction device of this embodiment is used to implement the aforementioned thermal power equipment efficiency prediction method. Therefore, the specific implementation method of the thermal power equipment efficiency prediction device can be seen in the implementation example part of the thermal power equipment efficiency prediction method in the previous text. For example, the data acquisition module, the model initialization module, the fitness value calculation module, the fitness value update module, and the prediction module are respectively used to implement steps S101, S102, S103, S104 and S105 in the above-mentioned thermal power equipment efficiency prediction method. Therefore, its specific implementation method can refer to the description of the corresponding various parts of the implementation example, which will not be repeated here.

[0150] A specific embodiment of the present invention further provides a thermal power equipment efficiency prediction device, comprising: a memory for storing a computer program; and a processor for implementing the steps of the above-mentioned thermal power equipment efficiency prediction method when executing the computer program.

[0151] A specific embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for predicting the efficiency of thermal power equipment are implemented.

[0152] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0153] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0154] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0155] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0156] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A method for predicting thermal power equipment efficiency, characterized in that: include: Step 1: Obtain data on factors affecting desulfurization efficiency, establish training sets and test sets, and perform data dimensionality reduction; Step 2: Initialize the SVM model and use chaos to initialize ISOA population parameters to generate the initialization population; Step 3: Input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold and solve the fitness value; Step 4: Introduce the sine-cosine algorithm to improve the attack behavior of individual seagulls, and after the individual seagulls arrive at the new position, introduce the opposite solution of the inertia weight for each individual seagull, update the individual position of the seagull, and re-solve the fitness value. The formula for introducing the sine-cosine algorithm to improve the attack behavior of individual seagulls is expressed as: X attack (t)=X new (t)·x·y·z+X best (t) Among them, X(t) represents the current location of the seagull individual, X attack (t) represents the attack point of the seagull, rand∈[0,1] represents a random number, X best (t) represents the current optimal position of the seagull individual, X new (t) represents the new position that the individual arrives at, r represents the spiral radius of each spiral motion of the seagull individual, ρ∈[0,2π] represents the random angle, and u and v represent the related constants of the spiral motion; Step 5: When the algorithm iteration limit is reached or the optimal fitness value is obtained, the penalty factor and the optimal parameter set of the kernel function are output to the SVM model to obtain a trained SVM model, the test set is input into the trained SVM model, and the prediction result is output.

2. The method for predicting thermal power equipment efficiency according to claim 1, characterized in that: When the algorithm iteration upper limit is not reached or the optimal fitness value is not obtained, return to step 4.

3. The method for predicting thermal power equipment efficiency according to claim 1, characterized in that: The data dimension reduction process comprises: Performing sample standardization processing on the desulfurization efficiency influencing factor data to obtain a correlation coefficient matrix, eigenvalues ​​and eigenvectors of the samples; Calculating the contribution rate of each component of the desulfurization efficiency influencing factor data; The eigenvector of the corresponding principal component whose contribution rate is greater than the preset threshold is multiplied by the original data to obtain the characteristic matrix of the factors affecting the desulfurization efficiency after dimensionality reduction.

4. The method for predicting thermal power equipment efficiency according to claim 1, characterized in that: The method of using chaos to initialize ISOA population parameters and generating an initialized population includes: A standard SOA population is randomly generated, and the distribution of the standard SOA population is optimized by using a Sinusoidal chaotic mapping to obtain the initialization population.

5. The method for predicting thermal power equipment efficiency according to claim 1, characterized in that: After the seagulls arrive at the new position, the opposite solution of the inertia weight is introduced for each seagull, and the calculation formula for updating the position of the seagulls is: Among them, X renew (t) is the updated position of the individual seagull, UB and LB represent the upper and lower limits of the optimization parameters, T max represents the upper limit of iteration, and t represents the current iteration number.

6. A thermal power equipment efficiency prediction device, characterized in that: include: The data acquisition module is used to obtain the data of factors affecting desulfurization efficiency, establish training sets and test sets, and perform data dimension reduction processing; Model initialization module, used to initialize the SVM model and use chaos to initialize ISOA population parameters to generate the initialization population; A fitness value calculation module is used to input the training set, set the penalty factor and kernel function to be optimized as the coordinates of the search threshold, and solve the fitness value; The fitness value update module is used to introduce the sine-cosine algorithm to improve the attack behavior of individual seagulls, and after the individual seagulls arrive at the new position, introduce the opposite solution of the inertia weight for each individual seagull, update the individual position of the seagull, and re-solve the fitness value, wherein the formula for introducing the sine-cosine algorithm to improve the attack behavior of individual seagulls is expressed as: X attack (t)=X new (t)·x·y·z+X best (t) Among them, X(t) represents the current location of the seagull individual, X attack (t) represents the attack point of the seagull, rand∈[0,1] represents a random number, X best (t) represents the current optimal position of the seagull individual, X new (t) represents the new position that the individual arrives at, r represents the spiral radius of each spiral motion of the seagull individual, ρ∈[0,2π] represents the random angle, and u and v represent the related constants of the spiral motion; The prediction module is used to output the optimal parameter set of the penalty factor and the kernel function to the SVM model when the algorithm iteration upper limit is reached or the optimal fitness value is obtained, so as to obtain a trained SVM model, input the test set into the trained SVM model, and output the prediction result.

7. A thermal power equipment efficiency prediction device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of a method for predicting the efficiency of thermal power equipment as described in any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for predicting the efficiency of thermal power equipment as claimed in any one of claims 1 to 5 are implemented.

Images