Gas turbine high load and low load switching control system, method, and storage medium

Abstract

The embodiment of the application provides a kind of gas turbine high load and low load switching control system, method and storage medium, solve the problem that existing method cannot accurately predict the load of thermal power generating set by combining the historical load of thermal power generating set, leading to the slow response speed of high load and low load switching, method includes: obtaining gas turbine operating parameter, executing load prediction model, outputting gas turbine load prediction curve, adjusting the load prediction curve by optimizing gas turbine operating parameter, obtain adjustment prediction curve, generate high load, low load switching strategy;The load prediction model is constructed by combining multi-hidden layer feedforward neural network and LSTM model in the application, the load prediction model can effectively use the time series characteristics of gas turbine operating parameters to accurately predict the load, and the load prediction model has fast convergence speed and good robustness, improves the response speed of high load and low load switching, and ensures the efficiency of thermal power generating set.

Description

TECHNICAL FIELD

[0001] The embodiment of the application relates to the technical field of gas turbine control, and particularly relates to a gas turbine high-load and low-load switching control system and method and a storage medium. BACKGROUND

[0002] In the operation process of the gas turbine, with the change of the load, the switching of the combustion mode can occur, and the switching is crucial for maintaining the stable combustion of the gas turbine and meeting the emission standard, in order to realize the effective coordination between the gas turbine and the steam turbine, a suitable control strategy needs to be designed, and the actual load of the steam turbine needs to be considered to realize the balance of the load instruction.

[0003] Chinese patent CN107528349A discloses a thermal power generator set load variable speed regulation control method and system based on regional new energy real-time power generation load, according to the new energy power generation capacity and the power grid operation demand in the region, the switching threshold of the thermal power generator set load variable speed regulation is set, when the new energy power generation load in the region is greater than the threshold, the associated thermal power generator set automatically completes the disturbance-free switching of the load regulation from low-speed regulation to high-speed regulation; when the new energy power generation load in the region is less than the threshold, the associated thermal power generator set automatically completes the disturbance-free switching of the load regulation from high-speed regulation to low-speed regulation; but the existing method cannot accurately predict the thermal power generator set load in combination with the historical load of the thermal power generator set, resulting in slow response speed of the high-load and low-load switching, affecting the efficiency of the thermal power generator set, and the gas turbine high-load and low-load switching control system, method and storage medium are proposed for the above problems. SUMMARY

[0004] The embodiment of the application aims to at least solve one of the technical problems in the prior art, and provides a gas turbine high-load and low-load switching control system, method and storage medium.

[0005] In a first aspect, the embodiment of the application provides a gas turbine high-load and low-load switching control method, which comprises the following steps:

[0006] obtaining a gas turbine working parameter; wherein the gas turbine working parameter comprises a unit temperature, a pressure, a rotating speed, a power output, a fuel flow, a vibration, an environment and a load parameter;

[0007] a load prediction model is constructed in combination with a multi-hidden layer feedforward neural network and an LSTM model, the gas turbine working parameter is taken as input, the load prediction model is executed, and a gas turbine load prediction curve is output;

[0008] The load prediction curve is obtained, the load prediction curve is decomposed into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, a load switching model is established by using a Boltzmann machine algorithm, the load switching model takes time-sharing energy consumption and power output as decision variables, the load prediction curve is adjusted by optimizing gas turbine working parameters to obtain an adjusted prediction curve.

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

[0010] An adjusted prediction curve is obtained, time periods are divided for the high load and the low load of the gas turbine based on the adjusted prediction curve as a constraint, and a high load and low load switching strategy is generated.

[0011] In some possible embodiments, the method of constructing a load prediction model by combining a multi-hidden-layer feedforward neural network and an LSTM model specifically comprises:

[0012] The LSTM model is used as an initial model of the load prediction model; wherein the initial model is composed of an autoencoder, two groups of LSTM layers, and a full connection layer, the LSTM layer includes three groups of LSTM blocks, and a multi-hidden-layer feedforward neural network is introduced between the autoencoder and the LSTM layer in the initial model;

[0013] The working parameters of the gas turbine are obtained, and the working parameters of the gas turbine are standardized processed;

[0014] In the standardization processing, the working parameters of the gas turbine in a day are standardized processed by the following formula:

[0015]

[0016] wherein, x i is the standardized working parameter of the gas turbine, r i is the input working parameter of the gas turbine, max{r i} and min{r i} are the maximum parameter and the minimum parameter of the working parameter of the gas turbine, l is the number of parameter types, and a is the parameter weight;

[0017] The standardized working parameters of the gas turbine are obtained, the bagging method is used to divide the working parameters of the gas turbine into a training set and a test set, and the proportion of the training set and the test set is 80% and 20%;

[0018] The training set is input into the autoencoder, and the autoencoder performs one-hot encoding on the training samples in the training set;

[0019] The activation function and the loss function of the multi-hidden-layer feedforward neural network are defined, the training samples are iteratively trained by combining the loss function with the gradient descent method to generate the output of the multi-hidden-layer feedforward neural network, wherein the activation function of the multi-hidden-layer feedforward neural network is the ReLU activation function, and the loss function is the MAPE loss function;

[0020] The number of neural network nodes of the LSTM layer is set to 100, the initial learning rate is set to 0.01, the training samples are trained by using the LSTM layer, and the output of the LSTM layer is obtained;

[0021] The full connection layer splices and fuses the output samples of the LSTM layer, and realizes parameter updating of the load prediction model to convergence;

[0022] The test set is obtained, the test set is taken as input, the load prediction model is executed, the predicted value of the test sample is obtained, and the MAPE value between the predicted value and the true value is calculated;

[0023] The MAPE value between the predicted value and the true value is divided into 16 equal parts, the proportion of the number of test samples in each interval is counted, and the PSI value of the load prediction model is calculated;

[0024] The PSI value is calculated by the following formula:

[0025]

[0026] Wherein, Ac d , Ec d are the probability values of the dth interval in the training samples and the test samples, respectively.

[0027] It is judged whether the PSI value of the load prediction model exceeds the preset probability threshold value, and if the preset probability threshold value is exceeded, the load prediction model is output.

[0028] In some possible embodiments, the method for executing the load prediction model and outputting the gas turbine load prediction curve specifically comprises:

[0029] The working parameters of the gas turbine are obtained, and the working parameters of the gas turbine are standardized to obtain a standardized set;

[0030] The standardized set is processed by dimension reduction to obtain a data dimension reduction set;

[0031] The data dimension reduction set is obtained, the data dimension reduction set is one-hot encoded by the autoencoder to obtain a data encoding set;

[0032] The data encoding set is loaded, and the data encoding set is divided into at least one group of initial prediction centers by the multi-hidden-layer feedforward neural network;

[0033] The association degree of the data coding set and the initial prediction center is calculated based on the FCM algorithm, wherein the association degree of the data coding set is calculated by the following formula:

[0034]

[0035] wherein m is the number of prediction centers, x b represents the input data coding set, x g represents the association degree of the data coding set, represents the mean of all prediction centers;

[0036] The association degree of the data coding set is obtained, and the prediction center to which the data coding set belongs is updated based on the association degree;

[0037] The posterior expectation variable of the data in the data coding set corresponding to the next period is updated based on the prediction center to which the data coding set belongs;

[0038]

[0039] wherein u i represents the posterior expectation variable of the next period, θ m represents the input representation of the prediction center, and q represents the prediction center weight penalty parameter;

[0040]

[0041] wherein q represents the prediction center weight penalty parameter, and Q represents the prediction center weight penalty constant.

[0042] In some possible embodiments, the method for executing the load prediction model and outputting the gas turbine load prediction curve further comprises the following steps:

[0043] The posterior expectation variable of the data in the data coding set corresponding to the next period is obtained, and the kernel matrix of the posterior expectation variable is calculated based on the covariance function;

[0044] The confidence upper bound, the confidence lower bound, and the expected candidate point defined by the confidence upper bound and the confidence lower bound of the kernel matrix are evaluated by using the obtaining function;

[0045] The expected candidate point defined by the confidence upper bound and the confidence lower bound is loaded, and the prediction load value associated with the expected candidate point is output by the LSTM layer;

[0046] The full connection layer integrates at least one group of prediction load values, determines the load state of the prediction load value based on the prediction load value, and outputs the gas turbine load prediction curve.

[0047] In some possible embodiments, the method for establishing the load switching model by using the Boltzmann machine algorithm comprises the following steps:

[0048] The load prediction curve is obtained, and the load prediction curve is decomposed by using an empirical mode decomposition algorithm to obtain a curve decomposition signal;

[0049] The empirical mode decomposition algorithm is represented as:

[0050]

[0051] J represents the load prediction curve decomposition signal, ω represents a center frequency of the load prediction curve interval frequency domain, j represents a noise tolerance of the empirical mode decomposition algorithm, and τ represents load prediction curve input data. i+t j i

[0052] The curve decomposition signal is obtained, and a load switching model is established by using a Boltzmann machine algorithm. The objective function of the load switching model is defined as:

[0053]

[0054] The load switching model outputs a load switching value, J(Q) represents a power output constraint, f(j) represents a time-sharing energy consumption constraint, and λ represents a switching coefficient of the load switching model.

[0055] In some possible embodiments, the method for adjusting the load prediction curve by optimizing the working parameters of the gas turbine specifically comprises:

[0056] The load prediction curve is obtained, and the load prediction curve is decomposed by using an empirical mode decomposition algorithm to obtain a curve decomposition signal;

[0057] The ant colony fitness and the convergence factor are calculated in combination with the Boltzmann machine function.

[0058] In some possible embodiments, the method for adjusting the load prediction curve by optimizing the working parameters of the gas turbine specifically further comprises:

[0059] The load prediction curve is adjusted by taking the time-sharing energy consumption and the power output as decision variables and in combination with the ant colony fitness and the convergence factor, to obtain an adjusted prediction curve.

[0060] In a second aspect, the embodiments of the present application provide a gas turbine high-load and low-load switching control system for implementing the gas turbine high-load and low-load switching control method as described above. The gas turbine high-load and low-load switching control system specifically comprises:

[0061] A data acquisition module is configured to obtain the working parameters of the gas turbine. The working parameters of the gas turbine include unit temperature, pressure, rotating speed, power output, fuel flow, vibration, environment, and load parameters.​​​​

[0062] The load prediction module combines a multi-hidden layer feedforward neural network and an LSTM model to construct a load prediction model, takes the working parameters of the gas turbine as input, executes the load prediction model, and outputs a load prediction curve of the gas turbine.

[0063] The load adjustment module is configured to obtain the load prediction curve, decompose the load prediction curve into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, establish a load switching model by using a Boltzmann machine algorithm, take the time-sharing energy consumption and power output as decision variables, adjust the load prediction curve by optimizing the working parameters of the gas turbine, and obtain an adjusted prediction curve.

[0064] The strategy output module is configured to obtain the adjusted prediction curve, divide the high load and the low load into time periods based on the adjusted prediction curve as a constraint, and generate a high load and low load switching strategy.

[0065] In a third aspect, a computer readable storage medium storing computer program instructions is provided, and the computer program instructions are executed to implement the gas turbine high load and low load switching control method as described above.

[0066] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0067] In the embodiments of the present application, the load prediction model is constructed by combining a multi-hidden layer feedforward neural network and an LSTM model, which can effectively utilize the time series characteristics of the working parameters of the gas turbine to accurately predict the load, and has a fast convergence speed and good robustness, thereby improving the high load and low load switching response speed, ensuring the efficiency of the thermal power generator unit, and overcoming the problem that the existing method cannot accurately predict the load of the thermal power generator unit in combination with the historical load of the thermal power generator unit, resulting in a slow high load and low load switching response speed and affecting the efficiency of the thermal power generator unit.

[0068] In the embodiments of the present application, the load prediction model and its training method are provided, which have the advantages of a multi-hidden layer feedforward neural network and an LSTM model, thereby being capable of learning the complex nonlinear relationship in the input data by using a multi-layer structure, effectively capturing long-term dependencies, improving the accuracy and efficiency of prediction, and enhancing the generalization ability and adaptability of the model.

[0069] In the embodiments of the present application, the load switching model is provided, which is built by using a Boltzmann machine algorithm, so that the load switching model takes the time-sharing energy consumption and power output as constraint conditions, ensures the efficiency and response speed during the high load and low load switching of the gas turbine, and also ensures the reduction of energy consumption and the stability of power output during the high load and low load switching. Attached Figure Description

[0070] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0071] Figure 1 This is a schematic diagram illustrating the implementation process of the gas turbine high-load and low-load switching control method provided in this embodiment of the invention;

[0072] Figure 2 This is a schematic diagram illustrating the implementation process of the method for constructing a load prediction model by combining a multi-hidden-layer feedforward neural network and an LSTM model, as provided in an embodiment of the present invention.

[0073] Figure 3 This is a schematic diagram illustrating the implementation process of the method for executing a load prediction model and outputting a gas turbine load prediction curve, provided in an embodiment of the present invention.

[0074] Figure 4 This is a schematic diagram illustrating the implementation process of the method for establishing a load switching model using the Boltzmann machine algorithm provided in an embodiment of the present invention;

[0075] Figure 5 This is a schematic diagram illustrating the implementation process of the method for adjusting the load prediction curve by optimizing the operating parameters of a gas turbine, as provided in an embodiment of the present invention.

[0076] Figure 6 This is a schematic diagram of the structure of the gas turbine high-load and low-load switching control system provided in an embodiment of the present invention. Detailed Implementation

[0077] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0078] Unless otherwise defined, technical terms and scientific terms used in the embodiments of the present application shall have the meanings that are commonly understood by one of ordinary skill in the art to which this application belongs. The terms "comprise", "include", "contain" and / or "have" used in the embodiments of the present application are intended to be construed as inclusive, unless otherwise specified. The terms "first", "second" and the like specified herein do not limit the scope of the application, but are used to distinguish one element from another. The terms "a", "an" and "the" used in the embodiments of the present application are used in the sense of "one or more", unless otherwise specified. The terms "comprise", "comprising", "include", "including" and "have" are used herein to mean including, but not limited to. The terms "exemplary", "for example" and "e.g." are used to provide examples, and are not used in a restrictive sense to design away from other embodiments. The terms "plurality" and "a plurality" as used herein mean two or more, unless otherwise specified.

[0079] Unless otherwise specified, the relative arrangement of components and steps, the numerical expressions, and numerical values set forth in the embodiments described herein are not limitations on the scope of the application. It should be understood that the dimensions of the various parts shown in the drawings are not necessarily to scale, and that the techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail, but should be considered as part of the description of the application. In all examples shown and discussed herein, any specific other examples can have different values. It should be noted that like symbols and letters in the drawings represent like items, and thus, once an item is defined in one drawing, it need not be discussed further in subsequent drawings.

[0080] In the description of the embodiments of the present application, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. The illustrative expressions of the above terms in the embodiments of the present application are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the embodiments of the present application and the features of the different embodiments or examples, without contradiction.

[0081] In the following, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and are not the whole embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0082] The existing method cannot combine the historical load of the thermal power generating unit to accurately predict the load of the thermal power generating unit, resulting in slow response speed of high load and low load switching, affecting the efficiency of the thermal power generating unit. In view of the above problems, the gas turbine high load and low load switching control system, method and storage medium are proposed. In short, when the method is executed, first, the working parameters of the gas turbine are obtained, a load prediction model is constructed by combining a multi-hidden layer feedforward neural network and an LSTM model, the working parameters of the gas turbine are input, the load prediction model is executed, and a load prediction curve of the gas turbine is output. Then, the load prediction curve is decomposed into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, and a load switching model is established by using a Boltzmann machine algorithm. Finally, the load switching model takes time-sharing energy consumption and power output as decision variables, adjusts the load prediction curve by optimizing the working parameters of the gas turbine, and obtains an adjusted prediction curve. In the embodiment of the present application, the load prediction model can effectively utilize the time series characteristics of the working parameters of the gas turbine to accurately predict the load, and the load prediction model has fast convergence speed and good robustness, which improves the high load and low load switching response speed, ensures the efficiency of the thermal power generating unit, and overcomes the problem that the existing method cannot combine the historical load of the thermal power generating unit to accurately predict the load of the thermal power generating unit, resulting in slow response speed of high load and low load switching, affecting the efficiency of the thermal power generating unit.

[0083] The embodiment of the present application provides a gas turbine high load and low load switching control method, Figure 1 The implementation process schematic diagram of the gas turbine high load and low load switching control method is shown, and the gas turbine high load and low load switching control method specifically includes:

[0084] Step S10, obtaining the working parameters of the gas turbine.

[0085] It should be noted that the working parameters of the gas turbine include but are not limited to unit temperature, pressure, speed, power output, fuel flow, vibration, environment, load parameters, and the working parameters of the gas turbine are acquired by temperature sensors, flow sensors, pressure sensors and vibration sensors based on a preset acquisition period. The preset acquisition period can be 1s-10min.

[0086] Step S20, constructing a load prediction model by combining a multi-hidden layer feedforward neural network and an LSTM model, inputting the working parameters of the gas turbine, executing the load prediction model, and outputting a load prediction curve of the gas turbine.

[0087] Step S30, obtain the load prediction curve, decompose the load prediction curve into at least one group of intrinsic mode function based on the empirical mode decomposition algorithm, and establish a load switching model by using the Boltzmann machine algorithm, wherein the load switching model takes the time-sharing energy consumption and power output as the decision variable, adjusts the load prediction curve by optimizing the working parameter of the gas turbine, and obtains an adjusted prediction curve.

[0088] Step S40, obtain the adjusted prediction curve, divide the high load and low load of the gas turbine into time periods based on the adjusted prediction curve as a constraint, and generate a high load and low load switching strategy.

[0089] In the embodiment of the present application, the load prediction model is constructed by combining the multi-hidden layer feedforward neural network and the LSTM model, which can effectively utilize the time sequence characteristics of the working parameters of the gas turbine for accurate load prediction, has a fast convergence speed and good robustness, improves the high load and low load switching response speed, ensures the efficiency of the thermal power generator unit, and overcomes the problem that the existing method cannot accurately predict the load of the thermal power generator unit by combining the historical load of the thermal power generator unit, resulting in a slow high load and low load switching response speed and affecting the efficiency of the thermal power generator unit.

[0090] The embodiment of the present application provides a method for constructing a load prediction model by combining a multi-hidden layer feedforward neural network and an LSTM model, Figure 2 The implementation process schematic diagram of the method for constructing a load prediction model by combining a multi-hidden layer feedforward neural network and an LSTM model is shown, and the method for constructing a load prediction model by combining a multi-hidden layer feedforward neural network and an LSTM model specifically comprises:

[0091] Step S101, taking the LSTM model as an initial model of the load prediction model, the initial model is composed of an autoencoder, two groups of LSTM layers and a full connection layer, the LSTM layer includes three groups of LSTM blocks, and a multi-hidden layer feedforward neural network is introduced between the autoencoder and the LSTM layer in the initial model.

[0092] Step S102, obtaining the working parameters of the gas turbine, and standardizing the working parameters of the gas turbine.

[0093] In the standardization process, the working parameters of the gas turbine in a day are standardized by the following formula:

[0094]

[0095] wherein, x i is the standardized working parameters of the gas turbine, r i is the input working parameters of the gas turbine, max{r i}, min{r i} is the maximum parameter, the minimum parameter, l is the parameter type number, and a is the parameter weight.

[0096] In step S103, the standardized gas turbine operating parameters are obtained, and the bagging method is used to divide the gas turbine operating parameters into a training set and a test set, and the ratio of the training set and the test set is 80%, 20%.

[0097] In step S104, the training set is input into the autoencoder, and the autoencoder performs one-hot encoding on the training samples in the training set.

[0098] In step S105, the activation function and the loss function of the multi-hidden layer feedforward neural network are defined, the training samples are iteratively trained by the loss function combined with the gradient descent method, and the output of the multi-hidden layer feedforward neural network is generated, wherein the activation function of the multi-hidden layer feedforward neural network is the ReLU activation function, and the loss function is the MAPE loss function.

[0099] In step S106, the number of neural network nodes of the LSTM layer is set to 100, the initial learning rate is set to 0.01, the training samples are trained by the LSTM layer, and the LSTM layer output is obtained.

[0100] In step S107, the full connection layer splices and fuses the LSTM layer output samples, and the parameter update of the load prediction model is realized to convergence.

[0101] In the embodiment of the present application, the load prediction model and the training method thereof are provided, which has the advantages of multi-hidden layer feedforward neural network and LSTM model, so as to learn the complex nonlinear relationship in the input data by using multi-layer structure, and then effectively capture the long-term dependence relationship, thereby improving the prediction accuracy and efficiency, and enhancing the generalization ability and adaptability of the model.

[0102] In step S108, the test set is obtained, the test set is input, the load prediction model is executed, the predicted value of the test sample is obtained, and the MAPE value between the predicted value and the true value is calculated.

[0103] In step S109, the MAPE value between the predicted value and the true value is divided into 16 equal parts, the proportion of the number of test samples in each interval is counted, and the PSI value of the load prediction model is calculated.

[0104] The PSI value is calculated by the following formula:

[0105]

[0106] Wherein, Ac d , Ec dThe probability value of the dth interval in the training sample and the test sample, respectively.

[0107] In step S110, it is judged whether the PSI value of the load prediction model exceeds a preset probability threshold.

[0108] In step S111, if the preset probability threshold is exceeded, the load prediction model is output. If the preset probability threshold is not exceeded, step S108 is returned, and the iterative training of the model is continued.

[0109] In the embodiment, the PSI value of the load prediction model is 0.032, which is less than the preset probability threshold 0.5, indicating that the stability and robustness of the load prediction model trained by the embodiment are better.

[0110] The embodiment of the present application provides a method for executing a load prediction model and outputting a gas turbine load prediction curve, Figure 3 The implementation process schematic diagram of the method for executing a load prediction model and outputting a gas turbine load prediction curve is shown, and the method for executing a load prediction model and outputting a gas turbine load prediction curve specifically includes:

[0111] In step S201, the working parameters of the gas turbine are obtained, the working parameters of the gas turbine are standardized, and a standardized set is obtained.

[0112] In step S202, the standardized set is dimensionally reduced to obtain a data dimensionally reduced set.

[0113] It should be noted that the method for dimensionally reducing the standardized set includes but is not limited to local linear embedding and factor analysis method for dimensionally reducing, wherein the local linear embedding method is used for dimensionally reducing, and it is assumed that each data point and its neighbor are located on a local linear manifold. LLE maps high-dimensional data to a low-dimensional space by keeping the reconstruction weight between each data point and its neighbor unchanged.

[0114] In step S203, the data dimensionally reduced set is obtained, the data dimensionally reduced set is one-hot encoded by the autoencoder, and a data encoded set is obtained.

[0115] In the embodiment of the present application, the one-hot encoding of the data dimensionally reduced set by the autoencoder not only solves the problem of classification data processing, but also improves the performance of the model. At the same time, since the standardized set has been dimensionally reduced, the shortcomings of dimension increase and information loss existing in one-hot encoding can be overcome.

[0116] In step S204, the data encoded set is loaded, and the data encoded set is divided into at least one initial prediction center by the multi-hidden layer feedforward neural network.

[0117] In step S205, the association degree of the data encoded set and the initial prediction center is calculated based on the FCM algorithm, wherein the association degree of the data encoded set is calculated by the following formula;

[0118]

[0119] wherein m is the number of prediction centers, x b represents the input data encoding set, x g represents the relevance of the data encoding set, represents the mean of all prediction centers.

[0120] Step S206, the relevance of the data encoding set is obtained, and the prediction center to which the data encoding set belongs is updated based on the relevance.

[0121] Step S207, the next period posterior expectation variable corresponding to the data in the data encoding set is updated based on the prediction center to which the data encoding set belongs.

[0122]

[0123] wherein u i represents the next period posterior expectation variable, theta m represents the input representation of the prediction center, q represents the prediction center weight penalty parameter;

[0124]

[0125] wherein q represents the prediction center weight penalty parameter, Q represents the prediction center weight penalty constant, and in the embodiment, the prediction center weight penalty constant is 2-8.

[0126] Step S208, the next period posterior expectation variable corresponding to the data in the data encoding set is obtained, and the kernel matrix of the posterior expectation variable is calculated based on the covariance function.

[0127] Step S209, the confidence upper bound, the confidence lower bound of the kernel matrix and the expected candidate point defined by the confidence upper bound and the confidence lower bound are evaluated by using the obtaining function.

[0128] Step S210, the expected candidate point defined by the confidence upper bound and the confidence lower bound is loaded, and the prediction load value associated with the expected candidate point is output by the LSTM layer.

[0129] Step S211, at least one group of prediction load values are integrated by the full connection layer, the load state of the prediction load value is determined based on the prediction load value, and the gas turbine load prediction curve is output.

[0130] The embodiment of the present application provides a method for establishing a load switching model by using a Boltzmann machine algorithm, Figure 4 The method for establishing a load switching model by using a Boltzmann machine algorithm is shown in the implementation process schematic diagram, and the method specifically comprises the following steps:

[0131] Step S301: Obtain the load forecast curve, and use the empirical mode decomposition algorithm to decompose the load forecast curve to obtain the curve decomposition signal;

[0132] The empirical mode decomposition algorithm is expressed as follows:

[0133]

[0134] Among them, J i+t This represents the decomposed signal of the load forecast curve, ω j j represents the center frequency in the frequency domain of the load forecast curve interval. i τ represents the input data for the load forecast curve, and τ represents the noise tolerance of the empirical mode decomposition algorithm.

[0135] Step S302: Obtain the curve decomposition signal and establish a load switching model using the Boltzmann machine algorithm. The objective function of the load switching model is defined as:

[0136]

[0137] in, The load switching value is output for the load switching model, where J(Q) represents the power output constraint, f(j) represents the time-of-use energy consumption constraint, and λ represents the switching coefficient of the load switching model.

[0138] In this embodiment of the invention, a load switching model is provided. The load switching model is constructed using the Boltzmann machine algorithm, so that the load switching model is constrained by time-of-use energy consumption and power output, ensuring the efficiency and response speed of the gas turbine when switching between high and low loads, while also ensuring reduced energy consumption and more stable power output when switching between high and low loads.

[0139] This invention provides a method for adjusting the load prediction curve by optimizing gas turbine operating parameters. Figure 5 A schematic diagram illustrating the implementation process of the method for adjusting the load prediction curve by optimizing gas turbine operating parameters is shown. The method for adjusting the load prediction curve by optimizing gas turbine operating parameters specifically includes:

[0140] Step S401: Obtain the load forecast curve, initialize the nonlinear convergence factor associated with the load forecast curve in the load switching model, and generate the initial position of the ant colony.

[0141] Step S402: Calculate the ant colony fitness and convergence factor using the Boltzmann machine function.

[0142] Step S403: Using time-of-use energy consumption and power output as decision variables, and combining ant colony fitness and convergence factor, adjust the load prediction curve to obtain the adjusted prediction curve.

[0143] In another aspect, the embodiment of the present application also provides a gas turbine high-load and low-load switching control system, Figure 6 A structural schematic diagram of the gas turbine high-load and low-load switching control system is shown, and the gas turbine high-load and low-load switching control system specifically comprises:

[0144] The data acquisition module 100 acquires the gas turbine working parameters, wherein the gas turbine working parameters include unit temperature, pressure, rotating speed, power output, fuel flow, vibration, environment and load parameters.

[0145] The load prediction module 200 combines a multi-hidden layer feedforward neural network and an LSTM model to construct a load prediction model, takes the gas turbine working parameters as input, executes the load prediction model, and outputs a gas turbine load prediction curve.

[0146] The load adjustment module 300 is used to acquire the load prediction curve, decompose the load prediction curve into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, establish a load switching model by using a Boltzmann machine algorithm, take the time-sharing energy consumption and power output as decision variables, adjust the load prediction curve by optimizing the gas turbine working parameters, and obtain an adjusted prediction curve.

[0147] The strategy output module 400 is used to acquire the adjusted prediction curve, divide the time periods of the gas turbine high-load and low-load based on the adjusted prediction curve as a constraint, and generate a high-load and low-load switching strategy.

[0148] In the embodiment, the gas turbine high-load and low-load switching control system is provided, which is composed of the data acquisition module 100, the load prediction module 200, the load adjustment module 300 and the strategy output module 400. The data acquisition module 100, the load prediction module 200, the load adjustment module 300 and the strategy output module 400 cooperatively utilize the time sequence characteristics of the gas turbine working parameters to accurately predict the load, improve the high-load and low-load switching response speed, and ensure the efficiency of the thermal power generator unit.

[0149] In another aspect, the embodiment of the present application also provides a computer readable storage medium, which stores computer program instructions, and the computer program instructions are executed to realize the gas turbine high-load and low-load switching control method.

[0150] As another aspect of the present application, a computer device is also provided, which comprises a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to realize the method of any one of the above embodiments.

[0151] The memory, as a non-volatile computer readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the gas turbine high load and low load switching control method in the embodiment of the present application. The memory can include a program storage area and a data storage area, wherein the program storage area can store an operating system and application programs required by at least one function; the data storage area can store data created by the use of the gas turbine high load and low load switching control method, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory can optionally include a memory remotely arranged relative to the processor, and these remote memories can be connected to the local module through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0152] Finally, it should be noted that the computer readable storage medium (for example, the memory) herein can be a volatile memory or a non-volatile memory, or can include both volatile memory and non-volatile memory. As an example but not limitation, the non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. The volatile memory can include random access memory (RAM), which can serve as an external cache memory. As an example but not limitation, the RAM can be obtained in various forms, such as synchronous RAM (DRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The storage device of the disclosed aspect is intended to include but not limited to these and other suitable types of memory.

[0153] In summary, the present application provides a gas turbine high load and low load switching control system, method and storage medium. In the embodiment of the present application, the load prediction model is constructed by combining the multi-hidden layer feedforward neural network and the LSTM model. The load prediction model can effectively utilize the time series characteristics of the gas turbine operating parameters to accurately predict the load, and has fast convergence speed and good robustness. The high load and low load switching response speed is improved, the thermal power generator set efficiency is ensured, and the problem that the existing method cannot accurately predict the thermal power generator set load by combining the historical load of the thermal power generator set, resulting in slow high load and low load switching response speed and affecting the thermal power generator set efficiency is overcome.

[0154] It should be noted that, for the foregoing embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the order of the described actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily necessary for the present application.

[0155] In several embodiments provided by the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic; for example, the division of the above units is only a logical function division, and there can be another division manner in actual implementation; for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or communication connection between each entity can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or in other forms.

[0156] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. they can be located in one place, or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0157] It can be understood that the above implementation manners are only exemplary implementation manners adopted for illustrating the principles of the present application, and the present application is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present application, and these modifications and improvements are also considered to be within the protection scope of the present application.

Claims

1. A gas turbine high load and low load switching control method, characterized by, The gas turbine high-load and low-load switching control method comprises: Obtaining gas turbine operating parameters; wherein the gas turbine operating parameters comprise unit temperature, pressure, rotating speed, power output, fuel flow, vibration, environment, load parameters; A load prediction model is constructed by combining a multi-hidden layer feedforward neural network and an LSTM model, the gas turbine operating parameters are taken as inputs, the load prediction model is executed, and a gas turbine load prediction curve is output; Obtaining the load prediction curve, decomposing the load prediction curve into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, establishing a load switching model by using a Boltzmann machine algorithm, taking time-sharing energy consumption and power output as decision variables, adjusting the load prediction curve by optimizing the gas turbine operating parameters, and obtaining an adjusted prediction curve; The method for executing the load prediction model and outputting the gas turbine load prediction curve specifically comprises: Obtaining the gas turbine operating parameters, and performing standardization processing on the gas turbine operating parameters to obtain a standardized set; Performing dimension reduction processing on the standardized set to obtain a data dimension reduction set; Obtaining the data dimension reduction set, and performing one-hot encoding on the data dimension reduction set by using a self-encoder to obtain a data encoding set; Loading the data encoding set, and dividing the data encoding set into at least one group of initial prediction centers by using a multi-hidden layer feedforward neural network; Calculating the correlation degree of the data encoding set and the initial prediction centers based on an FCM algorithm, wherein the correlation degree of the data encoding set is calculated by the following formula: where m is the number of prediction centers, x b denotes the input data encoding set, x g denotes the relevance of the data encoding set, denotes the mean of all prediction centers; Obtaining the correlation degree of the data encoding set, and updating the prediction center to which the data encoding set belongs based on the correlation degree; Updating the posterior expectation variable corresponding to the next period of the data in the data encoding set based on the prediction center to which the data encoding set belongs. where u i represents the posterior expected variable of the next period, θ m represents the input representation of the prediction center, q represents the prediction center weight penalty parameter; Wherein, q represents a prediction center weight penalty parameter, and Q represents a prediction center weight penalty constant.

2. The gas turbine high load and low load switching control method according to claim 1, characterized by, The method further comprises: Obtaining the adjusted prediction curve, dividing the high-load and low-load periods based on the adjusted prediction curve as a constraint, and generating a high-load and low-load switching strategy.

3. The gas turbine high and low load switching control method according to claim 1, characterized by, The method for constructing the load prediction model by combining the multi-hidden layer feedforward neural network and the LSTM model specifically comprises: Taking the LSTM model as an initial model of the load prediction model; wherein the initial model is composed of a self-encoder, two groups of LSTM layers, and a full connection layer, the LSTM layer includes three groups of LSTM blocks, and the multi-hidden layer feedforward neural network is introduced between the self-encoder and the LSTM layer in the initial model; Obtaining the gas turbine operating parameters, and performing standardization processing on the gas turbine operating parameters; Wherein, during the standardization processing, the gas turbine operating parameters are standardized by the following formula in a day: wherein x i is the normalized gas turbine operating parameter, r i is the input gas turbine operating parameter, max{r i}, min{r i} are the maximum and minimum parameters of the gas turbine operating parameter set, l is the number of parameter types, and a is the parameter weight. Obtaining the gas turbine operating parameters after the standardization processing, and dividing the gas turbine operating parameters into a training set and a test set by using a bagging method, the proportion of the training set and the test set is 80%, 20%; Inputting the training set into the self-encoder, and performing one-hot encoding on the training samples in the training set by using the self-encoder; The activation function and the loss function of the multi-hidden-layer feedforward neural network are defined, the training samples are iteratively trained by the loss function combined with the gradient descent method, and the output of the multi-hidden-layer feedforward neural network is generated, wherein the activation function of the multi-hidden-layer feedforward neural network is the ReLU activation function, and the loss function is the MAPE loss function; The number of neural network nodes of the LSTM layer is set to 100, the initial learning rate is set to 0.01, the training samples are trained by using the LSTM layer, and the LSTM layer output is obtained; The full connection layer splices and fuses the LSTM layer output samples, and realizes parameter updating of the load prediction model to convergence; The test set is obtained, the load prediction model is executed by taking the test set as input, the predicted value of the test sample is obtained, and the MAPE value between the predicted value and the true value is calculated; The MAPE value between the predicted value and the true value is equally divided into 16 intervals, the proportion of the number of test samples in each interval is counted, and the PSI value of the load prediction model is calculated. The PSI value is calculated by the following formula: wherein Ac d , Ec d are the probability values of the dth interval in the training sample, the test sample, respectively; It is judged whether the PSI value of the load prediction model exceeds the preset probability threshold value, and if it exceeds the preset probability threshold value, the load prediction model is output.

4. The gas turbine high and low load switching control method according to claim 1, characterized by, The method for executing the load prediction model and outputting the gas turbine load prediction curve further comprises: Obtain the posteriori expectation variable corresponding to the next period in the data coding set, and calculate the kernel matrix of the posteriori expectation variable based on the covariance function; The confidence upper limit, the confidence lower limit and the expected candidate point defined by the confidence upper limit and the confidence lower limit of the kernel matrix are evaluated by using the obtained function; The expected candidate point defined by the confidence upper limit and the confidence lower limit is loaded, and the predicted load value associated with the expected candidate point is output by the LSTM layer; The full connection layer integrates at least one group of predicted load values, determines the load state of the predicted load values based on the predicted load values, and outputs the gas turbine load prediction curve.

5. The gas turbine high and low load switching control method according to any one of claims 1 to 4, characterized by, The method for establishing a load switching model by using the Boltzmann machine algorithm comprises: Obtain the load prediction curve, and decompose the load prediction curve by using the empirical mode decomposition algorithm to obtain a curve decomposition signal; The empirical mode decomposition algorithm is represented as: wherein J i+t denotes the load prediction curve decomposition signal, ω j denotes the load prediction curve interval frequency domain center frequency, j i denotes the load prediction curve input data, τ denotes the empirical mode decomposition algorithm noise tolerance; Obtain the curve decomposition signal, and establish a load switching model by using the Boltzmann machine algorithm, wherein the objective function of the load switching model is defined as: wherein, The load switching model outputs a load switching value, J(Q) represents a power output constraint, f(j) represents a time-of-use energy consumption constraint, and λ represents a switching coefficient of the load switching model.

6. The gas turbine high and low load switching control method according to claim 5, characterized by, The method for adjusting the load prediction curve by optimizing the working parameters of the gas turbine comprises: Obtain the load prediction curve, initialize the nonlinear convergence factor associated with the load prediction curve by the load switching model, and generate the initial position of the ant colony; The fitness and the convergence factor of the ant colony are calculated by using the Boltzmann machine function.

7. The gas turbine high and low load switching control method according to claim 6, characterized by, The method for adjusting the load prediction curve by optimizing the working parameters of the gas turbine further comprises: The time-sharing energy consumption and the power output are taken as decision variables, the load prediction curve is adjusted based on the fitness and the convergence factor of the ant colony, and an adjusted prediction curve is obtained.

8. A gas turbine high and low load switching control system for implementing the gas turbine high and low load switching control method according to any one of claims 1 to 7, characterized by The gas turbine high-load and low-load switching control system comprises: A data acquisition module is configured to obtain the working parameters of the gas turbine, wherein the working parameters of the gas turbine include unit temperature, pressure, speed, power output, fuel flow, vibration, environment, and load parameters. The load prediction module combines a multi-hidden-layer feedforward neural network and an LSTM model to construct a load prediction model, takes the working parameters of the gas turbine as input, executes the load prediction model, and outputs a load prediction curve of the gas turbine; The load adjustment module is configured to obtain the load prediction curve, decompose the load prediction curve into at least one group of intrinsic mode functions based on an empirical mode decomposition algorithm, establish a load switching model by using a Boltzmann machine algorithm, take the time-sharing energy consumption and power output as decision variables, adjust the load prediction curve by optimizing the working parameters of the gas turbine, and obtain an adjusted prediction curve; The strategy output module is configured to obtain the adjusted prediction curve, divide the high load and the low load into time periods based on the adjusted prediction curve as a constraint, and generate a high load and low load switching strategy.

9. A computer-readable storage medium, characterized in that, The computer program instructions are executed to implement the gas turbine high load and low load switching control method according to any one of claims 1 to 7.

Images