Load prediction adjustment method and device for combined cycle power generation system

Abstract

The embodiment of the invention provides a load prediction adjustment method and device for a combined cycle power generation system, belongs to the technical field of combined cycle power generation, and solves the problems that when unit data is processed by an existing method, an LSTM neural network model is poor in self-adaption to the unit data, and gradient dispersion cannot be avoided. The method comprises the steps of acquiring sequential working parameters of the combined cycle power generation system, pre-constructing a load prediction model, outputting a load curve of the combined cycle power generation system in a next period, and realizing distributed adjustment calculation of the combined cycle power generation system based on load output constraint and load adjustment constraint of the combined cycle power generation system. According to the method, the load prediction model is constructed in combination with the swarm intelligence algorithm and the back propagation neural network, and the load prediction model has the advantages of the swarm intelligence algorithm and the back propagation neural network on sequential working parameter processing, so that the load curve of the combined cycle power generation system in the next period can be accurately output; and a basis is provided for distributed regulation of the combined cycle power generation system.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of combined cycle power generation, and in particular to a load prediction and regulation method and device for a combined cycle power generation system. Background Art

[0002] The combined cycle power generation system (CCPP) combines the working principles of gas turbines and steam turbines, and improves the overall thermal efficiency by connecting two different thermodynamic cycles in series. This system usually includes a gas turbine whose exhaust gas is used to generate steam to drive a steam turbine, thereby achieving cascade utilization of energy. The gas turbine is the most critical part of the combined cycle system, and its design and manufacturing level directly affects the performance and efficiency of the entire system. At the same time, accurate load prediction of power generation equipment in the combined cycle power generation system helps to reduce unnecessary energy waste, such as reducing fuel input during low-peak demand periods, thereby reducing overall energy consumption.

[0003] Chinese patent CN113107626B discloses a combined cycle power generation unit load prediction method based on multivariable LSTM, the method comprising the steps of: data collection; data preprocessing; dividing one unit data into a training set and a test set, and setting the other unit data as a validation set; building a multivariable input LSTM neural network model; substituting the validation set data into the model for prediction, and using the combined cycle power as the prediction target of the unit load; evaluating the model based on two evaluation indicators: the loss value in model training and the root mean square error in prediction; generating a true value-prediction value curve; however, when the existing method LSTM neural network model processes unit data, the LSTM neural network model has poor adaptability to the unit data and cannot avoid gradient diffusion, and thus cannot retain longer effective memory information; in response to the above problems, we propose a load prediction and regulation method and system for a combined cycle power generation system. Summary of the invention

[0004] The embodiments of the present invention aim to solve at least one of the technical problems existing in the prior art, and provide a load prediction and regulation method and device for a combined cycle power generation system.

[0005] In a first aspect, an embodiment of the present invention provides a load forecasting and regulating method for a combined cycle power generation system, the load forecasting and regulating method for a combined cycle power generation system comprising:

[0006] Acquiring sequential operating parameters of a combined cycle power generation system; wherein the combined cycle power generation system includes a gas turbine unit, a steam turbine, a waste heat boiler, and a generator; performing normalization processing on the sequential operating parameters to obtain a normalized data set;

[0007] Acquire the normalized data set, pre-build a load forecasting model combining a swarm intelligence algorithm and a back propagation neural network, wherein the load forecasting model uses the back propagation neural network as an initial model, uses the normalized data set to iteratively train the load forecasting model, and outputs a trained load forecasting model;

[0008] The real-time working parameters within a day are collected and normalized to obtain a real-time data set. The load forecasting model is executed with the real-time data set as input to output a load curve of the combined cycle power generation system within the next cycle.

[0009] In some possible embodiments, the method further includes:

[0010] The load curve of the combined cycle power generation system is obtained, and the load output constraint and load regulation constraint of the combined cycle power generation system are defined in combination with the sequential working parameters, and the distributed regulation calculation of the combined cycle power generation system is realized based on the load output constraint and load regulation constraint of the combined cycle power generation system.

[0011] In some possible embodiments, the method for normalizing the timing working parameters specifically includes:

[0012] Loading the sequential working parameters, completing the missing values ​​of the sequential working parameters, and removing abnormal values;

[0013] The timing working parameters are normalized, and the normalization formula is as follows:

[0014]

[0015] Among them, x(t) is the normalized value after normalization, t is the input timing working parameter, and t max is the maximum value of the timing working parameter, t min It is the minimum value of the timing working parameter;

[0016] The hybrid enhancement method is used to perform label hybrid enhancement on the normalized value. The label hybrid enhancement formula is:

[0017] mix-batch x(t) =α×batch x(n) +(1-α)batch x(t-n) (2)

[0018] mix-batch y(t) =α×batch y(n) +(1-α)batch y(t-n) (3)

[0019] Among them, batch x(n)For a batch of data to be processed, batch x(t-n) Another batch of data to be processed, batch y(n) The labels after mixed enhancement of a batch of data, batch y(t-n) is the label of another batch of data after mixed enhancement, α is the mixed enhancement coefficient, satisfying the following relationship;

[0020]

[0021] Where t is the number of data, x(t) represents the variance mean of the normalized value, and λ is the mixing enhancement constant;

[0022] The normalized values ​​after hybrid enhancement are vectorized to obtain a normalized data set.

[0023] In some possible embodiments, the method of iteratively training a load forecasting model using a normalized data set specifically includes:

[0024] The back propagation neural network is used as the initial model, which consists of an input layer, a hidden layer, and an activation layer. A gating structure based on a swarm intelligence algorithm is introduced into the input layer of the initial model.

[0025] Freeze the hidden layer of the initial model, introduce the forget gate, input gate, and output gate into the hidden layer of the initial model, and hide the feature extraction candidate values ​​for the training samples in the hidden layer. The calculation formula of the feature extraction candidate values ​​is as follows:

[0026]

[0027] in, Extract candidate values ​​for hidden features of training samples, tanh(·) is the activation function, and U C is the weight matrix, b c represents the bias term;

[0028] Combined cycle power generation system graph topology construction, mining the correlation between normalized data set and combined cycle power generation system load curve through graph topology expression, and introducing adaptive matrix in combined cycle power generation system graph topology to realize dynamic change of graph topology;

[0029] Obtaining a normalized data set, dividing the normalized data set into a training set and a test set, and initializing hyperparameters of the initial model;

[0030] Take the training set as input, execute the initial model, iteratively update and learn the internal parameters of the model, and obtain a converged initial model;

[0031] Obtain a test set, input the test set into the converged initial model, and obtain the test prediction result;

[0032] ACE is used to measure the confidence of the test prediction results. If the model confidence meets the preset confidence threshold, the trained load forecasting model is output.

[0033] In some possible embodiments, the use of ACE to measure the confidence of the test prediction result specifically includes:

[0034] Obtain the test prediction result, define the prediction confidence interval corresponding to the test prediction result, traverse the prediction confidence interval, and extract the quantile level corresponding to the quantile curve serving as the upper bound and the lower bound within the prediction confidence interval;

[0035] Calculate the nominal coverage probability corresponding to the forecast confidence interval and the interval empirical coverage probability;

[0036] Among them, the nominal coverage probability and interval empirical coverage probability are calculated by the following formula:

[0037]

[0038] Among them, E NCP is the nominal coverage probability, E ECP is the interval empirical coverage probability, Respectively represent the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, Respectively represent the predicted values ​​corresponding to the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, x t+l Indicates the size of the prediction interval;

[0039] Obtain the nominal coverage probability and the interval empirical coverage probability, and calculate the confidence of the test prediction result based on the nominal coverage probability and the interval empirical coverage probability;

[0040]

[0041] Among them, E ACE (x) is the confidence of the test prediction result, are the true values ​​of the corresponding upper and lower bound predictions, respectively.

[0042] In some possible embodiments, the method of executing the load forecasting model to output a load curve of the combined cycle power generation system in the next cycle specifically includes:

[0043] Get real-time working parameters within the day;

[0044] Normalize the real-time working parameters within the day to obtain a real-time data set;

[0045] The real-time data set is input into the input layer of the load forecasting model. The gating structure of the input layer optimizes the next cycle combination of the real-time data set based on the swarm intelligence algorithm to obtain the data optimization set.

[0046] In some possible embodiments, the method of executing the load forecasting model to output a load curve of the combined cycle power generation system in the next cycle specifically further includes:

[0047] Obtain a data optimization set, input the data optimization set into a hidden layer, and extract features of the data optimization set in the hidden layer to obtain a hidden feature set;

[0048] The activation layer traverses the hidden feature set, calculates the predicted load value of the combined cycle power generation system corresponding to the hidden feature set based on the activation function, integrates at least one group of predicted load values ​​of the combined cycle power generation system, and outputs the load curve of the combined cycle power generation system in the next cycle.

[0049] In some possible embodiments, the method for implementing the distributed regulation calculation of the combined cycle power generation system based on the load output constraint and the load regulation constraint of the combined cycle power generation system specifically includes:

[0050] Obtaining a load curve of a combined cycle power generation system, and defining a load output constraint and a load regulation constraint based on the load curve of the combined cycle power generation system and sequential working parameters;

[0051] Among them, the load output constraint is expressed as:

[0052] P l,min ≤P l ≤P l,max (9)

[0053]

[0054] P l Represents load output constraint, P l,min , P l,max They represent the lower and upper limit constraints of load output, β up , β down They represent the penalty cost coefficients for frequency regulation up and down of the system units, and they represent the penalty cost coefficients for peak regulation up and down of the system units, respectively. up ,p down represents the mean value of the peak load cost coefficient;

[0055] The load regulation constraint is expressed as:

[0056] S l,min ≤S l ≤S l,max (11)

[0057] Among them, S l represents the load regulation constraint, S l,min , S l,max They represent the lower and upper constraints of load regulation respectively;

[0058] Calculate the distributed regulation difference of the combined cycle power generation system based on the combined cycle power generation system load output constraint and load regulation constraint;

[0059] The distributed regulation difference of the combined cycle power generation system is obtained, and the distributed regulation of the combined cycle power generation system is realized based on the distributed regulation difference of the combined cycle power generation system.

[0060] In a second aspect, an embodiment of the present invention provides a load forecasting and regulating device for a combined cycle power generation system, which is used to implement the load forecasting and regulating method for a combined cycle power generation system as described above. The load forecasting and regulating device for a combined cycle power generation system specifically includes:

[0061] A data acquisition module is used to obtain the sequential working parameters of the combined cycle power generation system;

[0062] The model training module is used to obtain a normalized data set and pre-build a load forecasting model that combines a swarm intelligence algorithm and a back propagation neural network. The load forecasting model uses the back propagation neural network as the initial model, uses the normalized data set to iteratively train the load forecasting model, and outputs the trained load forecasting model.

[0063] The load curve output module is used to collect real-time working parameters within the day, normalize the real-time working parameters within the day, obtain a real-time data set, use the real-time data set as input, execute the load forecasting model, and output the load curve of the combined cycle power generation system in the next cycle;

[0064] The load regulation module defines the load output constraints and load regulation constraints of the combined cycle power generation system in combination with the timing working parameters, and realizes the distributed regulation calculation of the combined cycle power generation system based on the load output constraints and load regulation constraints of the combined cycle power generation system.

[0065] In some possible embodiments, the data acquisition module specifically includes:

[0066] A missing value processing unit loads the time series working parameters, completes the missing values ​​of the time series working parameters, and removes abnormal values;

[0067] A data enhancement unit is used to perform normalization processing on the temporal working parameters and perform label mixed enhancement on the normalized values ​​using a mixed enhancement method;

[0068] The vectorized representation unit is used to perform vectorized processing on the normalized values ​​after hybrid enhancement to obtain a normalized data set.

[0069] Compared with the prior art, the embodiments of the present invention mainly have the following beneficial effects:

[0070] In the embodiment of the present invention, a load forecasting model is constructed by combining a swarm intelligence algorithm and a back propagation neural network. The load forecasting model combines the advantages of the swarm intelligence algorithm and the back propagation neural network in processing sequential working parameters, so that the load curve of the combined cycle power generation system in the next cycle can be accurately output, thereby providing a basis for the distributed regulation of the combined cycle power generation system. The problem that the LSTM neural network model in the existing method processes unit data is poorly adaptive to the unit data, cannot avoid gradient diffusion, and thus cannot retain longer effective memory information is overcome.

[0071] In the embodiment of the present invention, by normalizing the timing working parameters, the modeling and training of the load forecasting model can be facilitated, and the normalization is performed by using a hybrid enhancement method combined with vectorization processing, which can help the model integrate data from different sources, solve the problem of data heterogeneity, and achieve seamless connection of multi-source data, providing a more comprehensive information basis for the comprehensive analysis of the load forecasting model, and helping to discover the deep relationships and patterns between the data.

[0072] In an embodiment of the present invention, a normalized data set is used to iteratively train a load forecasting model, and the load forecasting model consists of an input layer, a hidden layer, and an activation layer. At the same time, a gating structure based on a swarm intelligence algorithm is introduced into the input layer, thereby facilitating the construction of a topology of a cycle power generation system graph, improving the efficiency and quality of load forecasting model data processing, and providing strong support for the analysis and decision-making of complex systems.

[0073] In the embodiment of the present invention, ACE is used to measure the confidence of the test prediction results, so that the model evaluation does not rely on the probability threshold of the classifier output. Therefore, the performance of the model can be evaluated more comprehensively and the trade-off between sensitivity and specificity can be well handled. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0075] Figure 1 It is a schematic diagram of the implementation flow of the load forecasting and regulating method for the combined cycle power generation system provided by the present invention;

[0076] Figure 2 It is a schematic diagram of the implementation flow of the method for normalizing timing working parameters provided by the present invention;

[0077] Figure 3It is a schematic diagram of the implementation process of the method for iteratively training a load prediction model using a normalized data set provided by the present invention;

[0078] Figure 4 It is a schematic diagram of the implementation flow of using ACE to measure the confidence of test prediction results provided by the present invention;

[0079] Figure 5 It is a schematic diagram of the implementation flow of the method for executing the load prediction model provided by the present invention and outputting the load curve of the combined cycle power generation system in the next cycle;

[0080] Figure 6 It is a schematic diagram of the implementation flow of the distributed regulation calculation method for a combined cycle power generation system based on the load output constraint and load regulation constraint of the combined cycle power generation system provided by the present invention;

[0081] Figure 7 It is a structural schematic diagram of the load prediction and regulation device of the combined cycle power generation system provided by the present invention. DETAILED DESCRIPTION

[0082] In order to enable those skilled in the art to better understand the technical solution 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 part of the embodiments of the present invention, rather than all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0083] Unless otherwise specified, the technical terms or scientific terms used in the embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. "Including" or "comprising" used in the embodiments of the present invention neither limit the shapes, numbers, steps, actions, operations, components, originals and / or their groups mentioned, nor exclude the appearance or addition of one or more other different shapes, numbers, steps, actions, operations, components, originals and / or their groups, or the addition of these. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number and order of the indicated technical features. Thus, the features defined as "first" and "second" can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0084] Unless otherwise specifically stated, the relative arrangement of the components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention. At the same time, it should be understood that for ease of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship, and the techniques, methods and devices known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the techniques, methods and devices shown should be considered as part of the authorized specification. In all examples shown and discussed here, any specific other examples may have different values. It should be noted that similar symbols and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0085] In the description of the embodiments of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in the embodiments of the present invention and the features of the different embodiments or examples, without contradiction.

[0086] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0087] When the existing method LSTM neural network model processes the unit data, the LSTM neural network model has poor adaptability to the unit data and cannot avoid gradient diffusion, and thus cannot retain longer effective memory information. In response to the above problems, we propose a load forecasting and regulation method and system for a combined cycle power generation system. In short, when the method is executed, the sequential working parameters of the combined cycle power generation system are first obtained, and a load forecasting model combining a swarm intelligence algorithm and a back propagation neural network is pre-built. Then, the real-time working parameters within the day are collected, and the real-time working parameters within the day are normalized to obtain a real-time data set. The real-time data set is used as input to execute the load forecasting model, and the load curve of the combined cycle power generation system in the next cycle is output. Finally, the distributed regulation calculation of the combined cycle power generation system is realized based on the load output constraint and load regulation constraint of the combined cycle power generation system. In the embodiment of the present invention, a load forecasting model is constructed in combination with a swarm intelligence algorithm and a back propagation neural network. The load forecasting model has the advantages of the swarm intelligence algorithm and the back propagation neural network in processing sequential working parameters, so that the load curve of the combined cycle power generation system in the next cycle can be accurately output, thereby providing a basis for the distributed regulation of the combined cycle power generation system. The present invention overcomes the problem that when the LSTM neural network model processes unit data, the LSTM neural network model has poor adaptability to unit data, cannot avoid gradient diffusion, and thus cannot retain longer effective memory information.

[0088] The embodiment of the present invention provides a load forecasting and regulating method for a combined cycle power generation system. Figure 1 The schematic diagram of the implementation flow of the load forecasting and regulating method of the combined cycle power generation system is shown. The load forecasting and regulating method of the combined cycle power generation system specifically includes:

[0089] Step S10, obtaining the sequential operating parameters of the combined cycle power generation system, normalizing the sequential operating parameters, and obtaining a normalized data set;

[0090] It should be noted that the combined cycle power generation system includes but is not limited to gas turbine units, steam turbines, waste heat boilers, and generators, and the timing working parameters include the output power, load, speed, vibration, temperature, load curve, peak load, load fluctuation, solar radiation, ambient temperature, wind speed / direction of the system equipment.

[0091] Step S20, obtaining a normalized data set, pre-building a load forecasting model that combines a swarm intelligence algorithm and a back propagation neural network, the load forecasting model uses the back propagation neural network as an initial model, uses the normalized data set to iteratively train the load forecasting model, and outputs the trained load forecasting model;

[0092] Step S30, collecting real-time working parameters within the day, normalizing the real-time working parameters within the day to obtain a real-time data set, using the real-time data set as input, executing a load forecasting model, and outputting a load curve of the combined cycle power generation system in the next period.

[0093] It should be noted that the normalization method for the intraday real-time working parameters is the same as the normalization method for the time series working parameters.

[0094] Step S40, obtaining the load curve of the combined cycle power generation system, defining the load output constraint and load regulation constraint of the combined cycle power generation system in combination with the sequential working parameters, and realizing the distributed regulation calculation of the combined cycle power generation system based on the load output constraint and load regulation constraint of the combined cycle power generation system.

[0095] In the embodiment of the present invention, a load forecasting model is constructed by combining a swarm intelligence algorithm and a back propagation neural network. The load forecasting model combines the advantages of the swarm intelligence algorithm and the back propagation neural network in processing sequential working parameters, so that the load curve of the combined cycle power generation system in the next cycle can be accurately output, thereby providing a basis for the distributed regulation of the combined cycle power generation system. The problem that the LSTM neural network model in the existing method processes unit data is poorly adaptive to the unit data, cannot avoid gradient diffusion, and thus cannot retain longer effective memory information is overcome.

[0096] The embodiment of the present invention provides a method for normalizing timing operation parameters. Figure 2 A schematic diagram of the implementation process of a method for normalizing sequential working parameters is shown. The method for normalizing sequential working parameters specifically includes:

[0097] Step S101, loading sequential working parameters, completing missing values ​​of sequential working parameters, and removing abnormal values.

[0098] It should be noted that the missing values ​​of time series working parameters can be completed by deletion, filling and modeling prediction, while the outliers can be eliminated by standard deviation method, interquartile range method, regression residual method or isolation forest method.

[0099] Step S102, normalization processing is performed on the timing working parameters, and the normalization processing formula is as follows:

[0100] The timing working parameters are normalized, and the normalization formula is as follows:

[0101]

[0102] Among them, x(t) is the normalized value after normalization, t is the input timing working parameter, and t max is the maximum value of the timing working parameter, t min It is the minimum value of timing working parameter.

[0103] Step S103, a hybrid enhancement method is used to perform label hybrid enhancement on the normalized value, and the label hybrid enhancement formula is:

[0104] mix-batch x(t) =α×batch x(n) +(1-α)batch x(t-n) (2)

[0105] mix-batch y(t) =α×batch y(n) +(1-α)batch y(t-n) (3)

[0106] Among them, batch x(n) For a batch of data to be processed, batch x(t-n) Another batch of data to be processed, batch y(n) The labels after mixed enhancement of a batch of data, batch y(t-n) is the label of another batch of data after mixed enhancement, α is the mixed enhancement coefficient, satisfying the following relationship;

[0107]

[0108] Among them, t is the number of data, represents the variance mean of the normalized value, λ is the hybrid enhancement constant, and in this embodiment, the hybrid enhancement constant is an integer between 1 and 6.

[0109] Step S104: vectorize the normalized values ​​after hybrid enhancement to obtain a normalized data set.

[0110] In the embodiment of the present invention, by normalizing the timing working parameters, the modeling and training of the load forecasting model can be facilitated, and the normalization is performed by using a hybrid enhancement method combined with vectorization processing, which can help the model integrate data from different sources, solve the problem of data heterogeneity, and achieve seamless connection of multi-source data, providing a more comprehensive information basis for the comprehensive analysis of the load forecasting model, and helping to discover the deep relationships and patterns between the data.

[0111] The embodiment of the present invention provides a method for iteratively training a load prediction model using a normalized data set. Figure 3The following is a schematic diagram of the implementation process of a method for iteratively training a load prediction model using a normalized data set. The method for iteratively training a load prediction model using a normalized data set specifically includes:

[0112] Step S201, using a back propagation neural network as an initial model, the initial model consists of an input layer, a hidden layer, and an activation layer, and introducing a gating structure based on a swarm intelligence algorithm into the input layer of the initial model.

[0113] Step S202, freeze the hidden layer of the initial model, introduce a forget gate, an input gate, and an output gate into the hidden layer of the initial model, and the hidden layer hides the feature extraction candidate value for the training sample. The feature extraction candidate value calculation formula is as follows:

[0114] C~ t =tanh(W C x(t)+U C h t-1 +b c )(5)

[0115] Among them, C~ t Extract candidate values ​​for hidden features of training samples, tanh(·) is the activation function, and U C is the weight matrix, b c Represents the bias term.

[0116] Step S203, the combined cycle power generation system graph topology is constructed, the correlation between the normalized data set and the combined cycle power generation system load curve is mined through the graph topology expression, and an adaptive matrix is ​​introduced into the combined cycle power generation system graph topology to realize the dynamic change of the graph topology.

[0117] Step S204, obtaining a normalized data set, dividing the normalized data set into a training set and a test set, and initializing the hyperparameters of the initial model.

[0118] Step S205, using the training set as input, executing the initial model, iteratively updating and learning the internal parameters of the model, and obtaining a converged initial model.

[0119] Step S206, obtaining a test set, inputting the test set into the converged initial model, and obtaining a test prediction result.

[0120] Step S207, using ACE to measure the confidence of the test prediction results, if the model confidence meets the preset confidence threshold, output the trained load forecasting model.

[0121] In an embodiment of the present invention, a normalized data set is used to iteratively train a load forecasting model, and the load forecasting model consists of an input layer, a hidden layer, and an activation layer. At the same time, a gating structure based on a swarm intelligence algorithm is introduced into the input layer, thereby facilitating the construction of a topology of a cycle power generation system graph, improving the efficiency and quality of load forecasting model data processing, and providing strong support for the analysis and decision-making of complex systems.

[0122] The embodiment of the present invention provides a method for using ACE to measure the confidence of test prediction results. Figure 4 The following is a schematic diagram of the implementation process of using ACE to measure the confidence of the test prediction results. The method of using ACE to measure the confidence of the test prediction results specifically includes:

[0123] Step S2071, obtain the test prediction result, define the prediction confidence interval corresponding to the test prediction result, traverse the prediction confidence interval, and extract the quantile level corresponding to the quantile curve serving as the upper bound and the lower bound within the prediction confidence interval.

[0124] Step S2072, calculating the nominal coverage probability and the interval empirical coverage probability corresponding to the prediction confidence interval;

[0125] Among them, the nominal coverage probability and interval empirical coverage probability are calculated by the following formula:

[0126]

[0127] Among them, E NCP is the nominal coverage probability, E ECP is the interval empirical coverage probability, Respectively represent the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, Respectively represent the predicted values ​​corresponding to the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, x t+l Indicates the size of the prediction interval.

[0128] Step S2073, obtaining the nominal coverage probability and the interval empirical coverage probability, and calculating the confidence of the test prediction result based on the nominal coverage probability and the interval empirical coverage probability;

[0129]

[0130] Among them, E ACE (x) is the confidence of the test prediction result, are the true values ​​of the corresponding upper and lower bound predictions, respectively.

[0131] In the embodiment of the present invention, ACE is used to measure the confidence of the test prediction results, so that the model evaluation does not rely on the probability threshold of the classifier output. Therefore, the performance of the model can be evaluated more comprehensively and the trade-off between sensitivity and specificity can be well handled.

[0132] The embodiment of the present invention provides a method for executing a load forecasting model and outputting a load curve of a combined cycle power generation system in the next cycle. Figure 5 The present invention shows a schematic diagram of the implementation flow of a method for executing a load prediction model and outputting a load curve of a combined cycle power generation system in the next cycle. The method for executing a load prediction model and outputting a load curve of a combined cycle power generation system in the next cycle specifically includes:

[0133] Step S301, obtaining real-time working parameters within the day.

[0134] Step S302, normalizing the real-time working parameters within the day to obtain a real-time data set.

[0135] Step S303, input the real-time data set into the input layer of the load forecasting model, and the gating structure of the input layer optimizes the next cycle combination of the real-time data set based on the swarm intelligence algorithm to obtain a data optimization set.

[0136] Step S304, obtaining a data optimization set, inputting the data optimization set into a hidden layer, and extracting features of the data optimization set by the hidden layer to obtain a hidden feature set.

[0137] Step S305, the activation layer traverses the hidden feature set, calculates the predicted load value of the combined cycle power generation system corresponding to the hidden feature set based on the activation function, integrates at least one group of predicted load values ​​of the combined cycle power generation system, and outputs the load curve of the combined cycle power generation system in the next period.

[0138] The embodiment of the present invention provides a method for realizing distributed regulation calculation of a combined cycle power generation system based on load output constraints and load regulation constraints of the combined cycle power generation system. Figure 6 A schematic diagram of the implementation flow of a method for implementing a distributed regulation calculation method for a combined cycle power generation system based on load output constraints and load regulation constraints of a combined cycle power generation system is shown. The method for implementing a distributed regulation calculation method for a combined cycle power generation system based on load output constraints and load regulation constraints of a combined cycle power generation system specifically includes:

[0139] Step S401, obtaining a load curve of a combined cycle power generation system, and defining a load output constraint and a load regulation constraint based on the load curve of the combined cycle power generation system and sequential working parameters;

[0140] Among them, the load output constraint is expressed as:

[0141] P l,min ≤Pl ≤P l,max (9)

[0142]

[0143] P l Represents load output constraint, P l,min , P l,max They represent the lower and upper limit constraints of load output, β up , β down They represent the penalty cost coefficients for frequency regulation up and down of the system units, and they represent the penalty cost coefficients for peak regulation up and down of the system units, respectively. up ,p down represents the mean value of the peak load cost coefficient;

[0144] The load regulation constraint is expressed as:

[0145] S l,min ≤S l ≤S l,max (11)

[0146] Among them, S l represents the load regulation constraint, S l,min , S l,max They represent the lower and upper constraints of load regulation respectively.

[0147] Step S402: calculating the distributed regulation difference of the combined cycle power generation system based on the load output constraint and the load regulation constraint of the combined cycle power generation system.

[0148] Step S403, obtaining the distributed regulation difference of the combined cycle power generation system, and implementing the distributed regulation of the combined cycle power generation system based on the distributed regulation difference of the combined cycle power generation system.

[0149] The embodiment of the present invention also provides a load prediction and regulation device for a combined cycle power generation system. Figure 7 The structure diagram of the load prediction and regulation device of the combined cycle power generation system is shown, and the load prediction and regulation device of the combined cycle power generation system specifically includes:

[0150] The data acquisition module 100 is used to obtain the sequential operating parameters of the combined cycle power generation system.

[0151] The model training module 200 is used to obtain a normalized data set and pre-build a load forecasting model that combines a swarm intelligence algorithm and a back propagation neural network. The load forecasting model uses a back propagation neural network as an initial model, uses a normalized data set to iteratively train the load forecasting model, and outputs a trained load forecasting model.

[0152] The load curve output module 300 is used to collect real-time working parameters within the day, normalize the real-time working parameters within the day, obtain a real-time data set, use the real-time data set as input, execute the load prediction model, and output the load curve of the combined cycle power generation system in the next period.

[0153] The load regulation module 400 defines the load output constraint and load regulation constraint of the combined cycle power generation system in combination with the sequential working parameters, and implements the distributed regulation calculation of the combined cycle power generation system based on the load output constraint and load regulation constraint of the combined cycle power generation system.

[0154] In an embodiment of the present invention, a load prediction and regulation device for a combined cycle power generation system is provided. The load prediction and regulation device for a combined cycle power generation system is composed of a data acquisition module 100, a model training module 200, a load curve output module 300, and a load regulation module 400. The data acquisition module 100, the model training module 200, the load curve output module 300, and the load regulation module 400 cooperate to build a load prediction model in combination with a swarm intelligence algorithm and a back propagation neural network. The load prediction model has the advantages of swarm intelligence algorithms and back propagation neural networks in processing time-series working parameters, so that the load curve of the combined cycle power generation system in the next cycle can be accurately output, thereby providing a basis for the distributed regulation of the combined cycle power generation system. The problem that the LSTM neural network model in the existing method processes unit data is poorly adaptive to the unit data, cannot avoid gradient diffusion, and thus cannot retain longer effective memory information is overcome.

[0155] In this embodiment, the data acquisition module 100 specifically includes:

[0156] The missing value processing unit 110 is used to load the time series working parameters, complete the missing values ​​of the time series working parameters, and remove abnormal values. The data enhancement unit 120 is used to perform normalization processing on the time series working parameters, and use the hybrid enhancement method to perform label hybrid enhancement on the normalized values. The vectorized representation unit 130 is used to vectorize the normalized values ​​after hybrid enhancement to obtain a normalized data set.

[0157] On the other hand, an embodiment of the present invention further provides a structural schematic diagram of a computer device, which includes a display screen, a memory, a processor and a computer program, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the load forecasting and regulation method of the combined cycle power generation system.

[0158] It is understandable that, in the preferred embodiment provided by the present invention, the computer device may also be a notebook computer, a personal digital assistant (PDA), a mobile phone or other device capable of communication.

[0159] In another aspect of the embodiments of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer program instructions, and the computer program instructions can be executed by a processor. When the computer program instructions are executed, the method of any one of the above embodiments is implemented.

[0160] Exemplarily, the computer program may be divided into one or more modules, one or more modules are stored in a memory and executed by a processor to complete the present invention. One or more modules may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in a terminal device. For example, the above-mentioned computer program may be divided into units or modules of the load forecasting and regulating device of the combined cycle power generation system provided by the above-mentioned various system embodiments.

[0161] Finally, it should be noted that the computer-readable storage medium (e.g., memory) herein may be a volatile memory or a nonvolatile memory, or may include both a volatile memory and a nonvolatile memory. As an example and not by way of limitation, a nonvolatile memory may include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), or a flash memory. A volatile memory may include a random access memory (RAM), which may act as an external cache memory. As an example and not by way of limitation, RAM may be obtained in a variety of forms, such as synchronous RAM (DRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The storage devices of the disclosed aspects are intended to include, but are not limited to, these and other suitable types of memory.

[0162] The various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure herein may be implemented or executed using the following components designed to perform the functions herein: a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination of these components. A general purpose processor may be a microprocessor, but alternatively, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP, and / or any other such configuration.

[0163] In summary, the present invention provides a load forecasting and regulating method and system for a combined cycle power generation system. In the embodiment of the present invention, a load forecasting model is constructed by combining a swarm intelligence algorithm and a back propagation neural network. The load forecasting model combines the advantages of the swarm intelligence algorithm and the back propagation neural network in processing sequential working parameters, so that the load curve of the combined cycle power generation system in the next cycle can be accurately output, thereby providing a basis for the distributed regulation of the combined cycle power generation system. The problem that the LSTM neural network model in the existing method processes unit data is poorly adaptive to the unit data, cannot avoid gradient diffusion, and thus cannot retain longer effective memory information is overcome.

[0164] It should be noted that, for the above-mentioned embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, some steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0165] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A load forecasting and regulating method for a combined cycle power generation system, characterized in that: The load forecasting and regulating method of the combined cycle power generation system comprises: Acquiring sequential operating parameters of a combined cycle power generation system; wherein the combined cycle power generation system includes a gas turbine unit, a steam turbine, a waste heat boiler, and a generator; performing normalization processing on the sequential operating parameters to obtain a normalized data set; Acquire the normalized data set, pre-build a load forecasting model combining a swarm intelligence algorithm and a back propagation neural network, wherein the load forecasting model uses the back propagation neural network as an initial model, uses the normalized data set to iteratively train the load forecasting model, and outputs a trained load forecasting model; The real-time working parameters within a day are collected and normalized to obtain a real-time data set. The load forecasting model is executed with the real-time data set as input to output a load curve of the combined cycle power generation system within the next cycle.

2. The load forecasting and regulating method for a combined cycle power generation system according to claim 1, characterized in that: The method further comprises: The load curve of the combined cycle power generation system is obtained, and the load output constraint and load regulation constraint of the combined cycle power generation system are defined in combination with the sequential working parameters, and the distributed regulation calculation of the combined cycle power generation system is realized based on the load output constraint and load regulation constraint of the combined cycle power generation system.

3. The load forecasting and regulating method for a combined cycle power generation system according to claim 1, characterized in that: The method for normalizing the timing working parameters specifically includes: Loading the sequential working parameters, completing the missing values ​​of the sequential working parameters, and removing abnormal values; The timing working parameters are normalized, and the normalization formula is as follows: Among them, x(t) is the normalized value after normalization, t is the input timing working parameter, and t max is the maximum value of the timing working parameter, t min It is the minimum value of the timing working parameter; The hybrid enhancement method is used to perform label hybrid enhancement on the normalized value. The label hybrid enhancement formula is: mix_batch x(t) =α×batch x(n) +(1-α)batch x(t-n) (2) mix_batch y(t) =α×batch y(n) +(1-α)batch y(t-n) (3) Among them, batch x(n) For a batch of data to be processed, batch x(t-n) Another batch of data to be processed, batch y(n) The labels after mixed enhancement of a batch of data, batch y(t-n) is the label of another batch of data after mixed enhancement, α is the mixed enhancement coefficient, satisfying the following relationship; Among them, t is the number of data, represents the variance mean of the normalized value, λ is the mixing enhancement constant; The normalized values ​​after hybrid enhancement are vectorized to obtain a normalized data set.

4. The load forecasting and regulating method for a combined cycle power generation system according to claim 2, characterized in that: The method of iteratively training a load forecasting model using a normalized data set specifically includes: The back propagation neural network is used as the initial model, which consists of an input layer, a hidden layer, and an activation layer. A gating structure based on a swarm intelligence algorithm is introduced into the input layer of the initial model. Freeze the hidden layer of the initial model, introduce the forget gate, input gate, and output gate into the hidden layer of the initial model, and hide the feature extraction candidate values ​​for the training samples in the hidden layer. The calculation formula of the feature extraction candidate values ​​is as follows: in, Extract candidate values ​​for hidden features of training samples, tanh(·) is the activation function, and U C is the weight matrix, b c represents the bias term; Combined cycle power generation system graph topology construction, mining the correlation between normalized data set and combined cycle power generation system load curve through graph topology expression, and introducing adaptive matrix in combined cycle power generation system graph topology to realize dynamic change of graph topology; Obtaining a normalized data set, dividing the normalized data set into a training set and a test set, and initializing hyperparameters of the initial model; Take the training set as input, execute the initial model, iteratively update and learn the internal parameters of the model, and obtain a converged initial model; Obtain a test set, input the test set into the converged initial model, and obtain the test prediction result; ACE is used to measure the confidence of the test prediction results. If the model confidence meets the preset confidence threshold, the trained load forecasting model is output.

5. The load forecasting and regulating method for a combined cycle power generation system according to claim 4, characterized in that: The ACE is used to measure the confidence of the test prediction results, specifically including: Obtain the test prediction result, define the prediction confidence interval corresponding to the test prediction result, traverse the prediction confidence interval, and extract the quantile level corresponding to the quantile curve serving as the upper bound and the lower bound within the prediction confidence interval; Calculate the nominal coverage probability corresponding to the forecast confidence interval and the interval empirical coverage probability; Among them, the nominal coverage probability and interval empirical coverage probability are calculated by the following formula: Among them, E NCP is the nominal coverage probability, E ECP is the interval empirical coverage probability, f Respectively represent the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, Respectively represent the predicted values ​​corresponding to the quantile levels corresponding to the quantile curves that serve as the upper and lower bounds, x t+l Indicates the size of the prediction interval; Obtain the nominal coverage probability and the interval empirical coverage probability, and calculate the confidence of the test prediction result based on the nominal coverage probability and the interval empirical coverage probability; Among them, E ACE (x) is the confidence of the test prediction result, are the true values ​​of the corresponding upper and lower bound predictions, respectively.

6. The load forecasting and regulating method for a combined cycle power generation system according to claim 1, characterized in that: The method of executing the load forecasting model to output the load curve of the combined cycle power generation system in the next cycle specifically includes: Get real-time working parameters within the day; Normalize the real-time working parameters within the day to obtain a real-time data set; The real-time data set is input into the input layer of the load forecasting model. The gating structure of the input layer optimizes the next cycle combination of the real-time data set based on the swarm intelligence algorithm to obtain the data optimization set.

7. The load forecasting and regulating method for a combined cycle power generation system according to claim 6, characterized in that: The method of executing the load prediction model to output the load curve of the combined cycle power generation system in the next cycle specifically includes: Obtain a data optimization set, input the data optimization set into a hidden layer, and extract features of the data optimization set in the hidden layer to obtain a hidden feature set; The activation layer traverses the hidden feature set, calculates the predicted load value of the combined cycle power generation system corresponding to the hidden feature set based on the activation function, integrates at least one group of predicted load values ​​of the combined cycle power generation system, and outputs the load curve of the combined cycle power generation system in the next cycle.

8. The load forecasting and regulating method for a combined cycle power generation system according to claim 7, characterized in that: The method for realizing distributed regulation calculation of a combined cycle power generation system based on load output constraints and load regulation constraints of the combined cycle power generation system specifically includes: Obtaining a load curve of a combined cycle power generation system, and defining a load output constraint and a load regulation constraint based on the load curve of the combined cycle power generation system and sequential working parameters; Among them, the load output constraint is expressed as: P l,min ≤P l ≤P l,max (9) P l Represents load output constraint, P l,min , P l,max They represent the lower and upper limit constraints of load output, β up , β down They represent the penalty cost coefficients for frequency regulation up and down of the system units, and they represent the penalty cost coefficients for peak regulation up and down of the system units, respectively. up ,p down represents the mean value of the peak load cost coefficient; The load regulation constraint is expressed as: S l,min ≤S l ≤S l,max (11) Among them, S l represents the load regulation constraint, S l,min , S l,max They represent the lower and upper constraints of load regulation respectively; Calculate the distributed regulation difference of the combined cycle power generation system based on the combined cycle power generation system load output constraint and load regulation constraint; The distributed regulation difference of the combined cycle power generation system is obtained, and the distributed regulation of the combined cycle power generation system is realized based on the distributed regulation difference of the combined cycle power generation system.

9. A load forecasting and regulating device for a combined cycle power generation system, used to implement the load forecasting and regulating method for a combined cycle power generation system as claimed in any one of claims 1 to 8, characterized in that: The load prediction and regulation device of the combined cycle power generation system specifically comprises: Data acquisition module, used to obtain the sequential working parameters of the combined cycle power generation system; The model training module is used to obtain a normalized data set and pre-build a load forecasting model that combines a swarm intelligence algorithm and a back propagation neural network. The load forecasting model uses the back propagation neural network as the initial model, uses the normalized data set to iteratively train the load forecasting model, and outputs the trained load forecasting model. The load curve output module is used to collect real-time working parameters within the day, normalize the real-time working parameters within the day, obtain a real-time data set, use the real-time data set as input, execute the load forecasting model, and output the load curve of the combined cycle power generation system in the next cycle; The load regulation module defines the load output constraints and load regulation constraints of the combined cycle power generation system in combination with the timing working parameters, and realizes the distributed regulation calculation of the combined cycle power generation system based on the load output constraints and load regulation constraints of the combined cycle power generation system.

10. The load forecasting and regulating device for a combined cycle power generation system according to claim 9, characterized in that: The data acquisition module specifically includes: A missing value processing unit loads the time series working parameters, completes the missing values ​​of the time series working parameters, and removes abnormal values; A data enhancement unit is used to perform normalization processing on the temporal working parameters and perform label mixed enhancement on the normalized values ​​using a mixed enhancement method; The vectorized representation unit is used to perform vectorized processing on the normalized values ​​after hybrid enhancement to obtain a normalized data set.

Images