Exhaust temperature control method and device for hydrogen gas turbine

Abstract

The invention discloses a method and device for controlling the exhaust temperature of a hydrogen gas turbine, and the method comprises the steps: collecting operation data which comprise historical operation data and real-time operation data; training is carried out according to the preprocessed operation data, a feedforward prediction model is obtained, and the feedforward prediction model comprises an adjustable adaptation layer; inputting the real-time operation data into the feed-forward prediction model to obtain a feed-forward prediction value, calculating to obtain an offset degree according to the feed-forward prediction value and the real-time operation data, and obtaining a feed-forward adjustment signal according to the offset degree; obtaining a compensation adjustment signal according to the real-time operation data and a preset signal; and weighting the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control quantity.

Description

Technical Field

[0001] The present invention relates to the technical field of data optimization management of gas turbine units, and particularly relates to a method and device for controlling the exhaust temperature of a hydrogen gas turbine. Background Art

[0002] As an efficient and clean power generation device, the combustion products of a hydrogen gas turbine are only water, which has the advantage of zero carbon emissions and is one of the important technical paths to achieve the "dual carbon" goal. There are significant differences in the combustion characteristics between hydrogen and natural gas. For example, hydrogen has a faster combustion speed and a higher flame temperature. Usually, the gas turbine monitors and controls the temperature at the outlet of the combustion chamber through the exhaust temperature of the gas turbine. The characteristics of hydrogen combustion increase the difficulty of gas turbine operation control.

[0003] Traditional gas turbine exhaust temperature control mainly adopts a feedback control strategy, that is, it adjusts according to the deviation between the actual value and the set value of the gas turbine exhaust temperature. There are still the following problems:

[0004] (1) Due to its active characteristics, hydrogen has a faster combustion speed. Since the temperature at the outlet of the gas turbine combustion chamber cannot be directly monitored, the exhaust temperature of the gas turbine is usually used for indirect monitoring and control. Therefore, the active characteristics of hydrogen combustion increase the difficulty of exhaust temperature control;

[0005] (2) Traditional feedback control (such as PI control) cannot predict disturbances in advance, resulting in poor control effects;

[0006] (3) Conventional feedforward control is usually based on a simplified mathematical model or empirical formula, which is difficult to accurately describe the complex dynamic characteristics of a hydrogen gas turbine. Under dynamic operating conditions, the model prediction error is large;

[0007] (4) Traditional control methods usually adopt fixed control parameters and cannot be adaptively adjusted according to the changes in the operating state of the unit.

[0008] In the patent document CN115492691A, a gas turbine control method, device and system are disclosed. Among them, the method includes: by receiving the measured value and predicted value of the measured data sent by the control platform, realizing inputting the measured value and predicted value of the measured data into a Kalman filter for optimal estimation to obtain the deviation of the measured data, and then according to the deviation of the measured data, using a machine learning algorithm to determine the deviation of the unmeasured data, and further sending the deviation of the measured data and the deviation of the unmeasured data to the online model of the gas turbine to correct the predicted values of the measured data and the predicted values of the unmeasured data output by the online model, to solve the problems that traditional gas turbine control methods for power generation mostly adopt feedback control, cannot predict dynamics, have poor adaptability to hydrogen combustion, and cannot adapt to the unit state by using fixed control parameters.

[0009] In the patent document CN116498440A, a method and device for controlling the IGV opening of a gas turbine based on generalized predictive control are disclosed, including establishing a transfer function model between the IGV opening of the gas turbine and the exhaust gas temperature, and identifying the initial parameters based on historical operation data; using the initial parameters to establish a prediction model for the IGV opening of the gas turbine and predicting the object output under different control quantities; introducing a step factor and performing rolling optimization on the control quantity to obtain the target control quantity as the IGV opening control command; online correcting the parameters of the prediction model by the recursive least squares method to improve the robustness of the control system, and to solve the problems that the traditional control methods for gas turbines used in power generation mostly adopt feedback control, there are problems such as inability to predict dynamics, poor adaptability to hydrogen combustion, and inability to adapt to the unit state by using fixed control parameters.

[0010] In summary, the above two existing patents have not solved the problems that the traditional control methods for gas turbines used in power generation mostly adopt feedback control, there are problems such as inability to predict dynamics, poor adaptability to hydrogen combustion, and inability to adapt to the unit state by using fixed control parameters. Summary of the Invention

[0011] Based on the above technical problems, the present invention proposes a method and device for controlling the exhaust gas temperature of a hydrogen gas turbine, which solves the problems that the traditional control methods for gas turbines used in power generation mostly adopt feedback control, there are problems such as inability to predict dynamics, poor adaptability to hydrogen combustion, and inability to adapt to the unit state by using fixed control parameters.

[0012] To achieve the above object, the present invention proposes a method for controlling the exhaust gas temperature of a hydrogen gas turbine.

[0013] A method for controlling the exhaust gas temperature of a hydrogen gas turbine includes:

[0014] Collecting operation data, where the operation data includes historical operation data and real-time operation data;

[0015] Training according to the preprocessed operation data to obtain a feedforward prediction model, where the feedforward prediction model includes an adjustable adaptation layer;

[0016] Inputting the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculating a bias degree according to the feedforward prediction value and the real-time operation data, and obtaining a feedforward adjustment signal according to the bias degree; obtaining a compensation adjustment signal according to the real-time operation data and a preset signal; weighting the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control quantity.

[0017] Further, it further includes:

[0018] Collect the real-time operation data through a sliding window to obtain an updated data set, and update the adaptable layer of the feedforward prediction model according to the updated data set and the feedforward prediction value through adaptive online gradient and low-rank adaptation.

[0019] Further, the operation data includes:

[0020] The compressor outlet pressure, the gas turbine outlet pressure, the opening value of the flow control device, the gas turbine power, the fuel calorific value, and the gas turbine exhaust temperature.

[0021] Further, the operation data further includes:

[0022] The ambient temperature, the ambient pressure, the ambient humidity, and / or the compressor outlet temperature.

[0023] Further, the flow control device includes:

[0024] The fuel valve and / or the inlet guide vane.

[0025] Further, collecting the operation data further includes:

[0026] Preprocessing the operation data.

[0027] Further, preprocessing the operation data includes:

[0028] Denosing, feature engineering, and standardizing the data;

[0029] Dividing it into a training set, a validation set, and a test set.

[0030] Further, training according to the preprocessed operation data includes:

[0031] Based on the operation data, construct a Transformer model, use the ambient temperature, the ambient pressure, the ambient humidity, the compressor outlet temperature, the compressor outlet pressure, the gas turbine outlet pressure, the gas turbine power, and the fuel calorific value as input data, and use the opening value of the flow control device and the gas turbine exhaust temperature as output targets, and train to obtain the feedforward prediction model.

[0032] Further, the feedforward prediction value includes:

[0033] The opening prediction value of the flow control device and the exhaust temperature prediction value of the gas turbine.

[0034] Further, calculating the bias degree according to the feedforward prediction value and the real-time operation data includes:

[0035] The bias degree is obtained by calculating the difference between the opening value of the flow control device in the real-time operation data and the predicted opening value.

[0036] Further, obtaining the feedforward adjustment signal according to the bias degree includes:

[0037] Calculating a deviation rate based on the bias degree and the predicted opening value, and determining the feedforward adjustment signal according to the deviation rate.

[0038] Further, obtaining a compensation adjustment signal according to the real-time operation data and a preset signal includes:

[0039] Obtaining a deviation signal according to the exhaust gas temperature of the gas turbine and the predicted exhaust gas temperature in the real-time operation data, and obtaining the compensation adjustment signal according to the deviation signal and the preset signal.

[0040] Further, updating the adaptable layer of the feedforward prediction model according to the updated data set and the feedforward prediction value through adaptive online gradient and low-rank adaptation includes:

[0041] Performing loss calculation according to the feedforward prediction value to obtain a loss value, obtaining a gradient according to the loss value, calculating a diagonal matrix according to the gradient, and updating the low-rank matrix in the adjustable adaptable layer according to the diagonal matrix and the gradient.

[0042] Further, collecting the real-time operation data by using a sliding window includes:

[0043] The window size is 25 days - 35 days, and the window period is 1h - 2h.

[0044] To achieve the above object, the present invention also proposes an exhaust gas temperature control device for a hydrogen gas turbine.

[0045] An exhaust gas temperature control device for a hydrogen gas turbine, comprising:

[0046] A collection module: used to collect operation data, and the operation data includes historical operation data and real-time operation data;

[0047] A collection module, used to collect operation data, the operation data includes historical operation data and real-time operation data, and preprocess the operation data;

[0048] A training module, used to train according to the preprocessed operation data to obtain a feedforward prediction model;

[0049] A control module, configured to input the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculate a bias degree based on the feedforward prediction value and the real-time operation data, and obtain a feedforward adjustment signal according to the bias degree; calculate and obtain a compensation adjustment signal through PID control based on the real-time operation data, a preset signal, and the feedforward prediction model; and weight the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control amount.

[0050] An update module, configured to collect the real-time operation data through a sliding window to obtain an updated data set, and update an adaptable layer of the feedforward prediction model through adaptive online gradient and low rank adaptation based on the updated data set and the feedforward prediction value.

[0051] Based on the above technical solutions, the present invention has at least the following beneficial effects:

[0052] 1. The present invention provides a method and device for controlling the exhaust gas temperature of a hydrogen gas turbine. Through a feedforward control strategy, it can predict disturbances in the system in advance, thereby reducing the impact of disturbances on the system performance, improving the stability and response speed of the system; combining the predicted results with traditional feedback control to obtain the final execution control amount can more accurately adjust the control amount, achieve optimized control of the system performance, improve the overall operation efficiency and economy of the system; the feedforward control strategy can effectively enhance the robustness of the system, enabling it to still maintain good performance and stability in the face of complex operating environments and uncertain factors, and reducing the probability of failures; the compensation adjustment strategy can effectively eliminate the steady-state error through the integral part of a traditional PID controller, ensuring high precision during long-term operation of the system.

[0053] 2. The present invention provides a method and device for controlling the exhaust gas temperature of a hydrogen gas turbine. By introducing an online learning and model update mechanism, it can monitor the operating state of the unit in real time and update the model according to the latest operation data, thereby ensuring that the model can always accurately reflect the actual operating conditions of the unit, improving the adaptability and optimization ability of the system; the online learning and model update mechanism realizes continuous improvement and optimization of the system performance by continuously obtaining real-time operation data. Through continuous learning and adjustment, the system can better cope with various changes and challenges, improve the overall operation efficiency and economy of the system, and at the same time can promptly detect and solve potential problems, improve the reliability and stability of the system, reduce the probability of failures, and ensure the long-term stable operation of the system.

[0054] 3. The present invention proposes a method and device for controlling the exhaust temperature of a hydrogen gas turbine. By applying an advanced Transformer model, it can accurately predict the exhaust temperature of the hydrogen gas turbine, thereby achieving precise control and optimization of the exhaust temperature. Utilizing the powerful modeling ability of the model, it can comprehensively consider various influencing factors, such as fuel flow rate, air flow rate, ambient temperature, etc., so as to achieve comprehensive optimization of the performance of the gas turbine, improve the overall operating efficiency and stability; by real-time monitoring the exhaust temperature and combining the feed-forward prediction value of the feed-forward prediction model, it can timely detect abnormal situations and make adjustments to ensure that the gas turbine operates in the best state and extend the equipment life. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings forming a part of this invention are used to provide a further understanding of the present invention. The schematic embodiments and descriptions thereof of the present invention are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0056] Figure 1 Shows a flowchart of a method for controlling the exhaust temperature of a hydrogen gas turbine in an embodiment;

[0057] Figure 2 Shows a schematic structural diagram of a device for controlling the exhaust temperature of a hydrogen gas turbine in an embodiment;

[0058] Figure 3 Shows a schematic diagram of a control process for the exhaust temperature of a hydrogen gas turbine in an embodiment;

[0059] Figure 4 Shows a schematic structural diagram of a product for controlling the exhaust temperature of a hydrogen gas turbine in an embodiment;

[0060] Figure 5 Shows a schematic structural diagram of an electronic device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0061] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0062] The following further describes the present invention in detail with reference to specific embodiments, and these embodiments should not be construed as limiting the scope claimed by the present invention.

[0063] Embodiment

[0064] To solve the problems that the traditional control methods for gas turbines used in power generation mostly adopt feedback control, there are problems such as inability to predict dynamics, poor adaptability to hydrogen combustion, and inability to adapt to the unit state with fixed control parameters, the present invention proposes a method and device for controlling the exhaust temperature of a hydrogen gas turbine.

[0065] To achieve the above object, the present invention also proposes a method for controlling the exhaust temperature of a hydrogen gas turbine.

[0066] As Figure 1 shows a method for controlling the exhaust temperature of a hydrogen gas turbine according to an embodiment of the present invention. The process mainly includes the following steps:

[0067] S101: Collect operation data, where the operation data includes historical operation data and real-time operation data.

[0068] Specifically, as Figure 3 shown in this embodiment, the flow rate of the gas turbine is jointly controlled by the fuel valve and the inlet guide vane (IGV). The flow rate devices are selected as the fuel valve and the inlet guide vane. In other embodiments, the flow rate of the gas turbine can also be controlled only by the fuel valve or only by the inlet guide vane.

[0069] Further, collect the historical operation data and real-time operation data of the gas turbine. The collected data includes the compressor outlet pressure, the gas turbine outlet pressure, the opening value of the fuel valve, the opening value of the inlet guide vane, the gas turbine power, the fuel calorific value, the gas turbine exhaust temperature, the ambient temperature, the ambient pressure, the ambient humidity, and the compressor outlet temperature. Take the opening value of the fuel valve, the opening value of the inlet guide vane, and the gas turbine exhaust temperature as the output targets, and take the compressor outlet pressure, the gas turbine outlet pressure, the gas turbine power, the fuel calorific value, the ambient temperature, the ambient pressure, the ambient humidity, and the compressor outlet temperature as the input data.

[0070] Further, denoise the collected operation data. In this embodiment, the Savitzky-Golay filter is used for denoising. First, use mean filtering to remove outliers and obvious noise points from the collected operation data, and then input the data into the Savitzky-Golay filter. Select the window size to be 5 and the polynomial order to be 2 to obtain the preprocessed historical operation data and real-time operation data.

[0071] Further, perform feature engineering on the collected operation data. The input historical sequence window length is n*50, where n is the time span unit of the input historical data in seconds, and 50 represents the sampling frequency of 50 Hz. The output window length is 0.02 seconds; the expression of each generated sample is:

[0072] X∈R L×d ,

[0073] The label is y∈R, where L is the length of the sequence, d is the number of features at each time step, initialize the sample list, and traverse the operation data to generate training samples.

[0074] Further, all the obtained training samples are subjected to data standardization. In this embodiment, the Z-Score method is adopted, and its expression is as follows:

[0075]

[0076] where x norm is the standardized data, x is the original data, μ is the average value of the training samples, and σ is the standard deviation of the training samples.

[0077] Further, the standardized announced sample data is divided. Specifically, in this embodiment, 70% of the sample data is randomly divided into the training set, 15% of the sample data is randomly divided into the validation set, and 15% of the sample data is randomly divided into the test set.

[0078] S102: Train according to the preprocessed operation data to obtain a feedforward prediction model, and the feedforward prediction model includes an adjustable adaptation layer.

[0079] Further, a Transform model is built. In this embodiment, the number of heads h in the multi-head self-attention layer of the model is set to 9, and the dimension of each head is 64; the dimension d of the hidden layer of the feedforward neural network is 36; the number of stacked layers is 6 encoder layers; the feature of the last time step output by the adjustable adaptation layer of the encoder is mapped to a scalar output through a fully connected layer:

[0080]

[0081] where y pred is the feedforward prediction value, W out is the weight matrix of the output layer, is the data input in the training set, b out is the bias of the output layer, which is used to adjust the baseline of y pred .

[0082] Further, two low-rank matrices A and B are respectively added to the adjustable adaptation layer of the model, so that the output of the model is expressed as:

[0083] W out = W out + α * AB,

[0084] where W out is the weight matrix of the output layer, α is the learning rate, A is the low-rank matrix A, and B is the low-rank matrix B.

[0085] Further, the model is trained according to the training set to obtain a feedforward prediction model.

[0086] Specifically, in this embodiment, the loss function is the mean square error (MSE), the optimizer is selected as AdamW with an initial learning rate of 1e-4 and a weight decay coefficient of 1e-5; the learning rate scheduling adopts cosine annealing, and the trained model is converted into the TensorRT format and embedded into the control system.

[0087] S103: Input the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculate a bias degree based on the feedforward prediction value and the real-time operation data, and obtain a feedforward adjustment signal according to the bias degree; obtain a compensation adjustment signal based on the real-time operation data, a preset signal, and the feedforward prediction model; weight the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control amount.

[0088] Further, as Figure 3 shown in, input the operation data of the gas turbine into the feedforward prediction model to obtain a predicted opening value of the fuel valve, a predicted opening value of the inlet guide vane, and a predicted exhaust temperature value. The expressions are as follows:

[0089] IGV pre (t) = TransMod_in(x(t)),

[0090] NGVal pre (t) = TransMod_in(x(t)),

[0091] where IGV pre (t) is the predicted opening value of the inlet guide vane, NGVal pre (t) is the predicted opening value of the fuel valve, and TransMod_in(x(t)) is the model input.

[0092] Further, calculate the difference between the predicted opening value of the fuel valve and the opening value of the fuel valve in the collected operation data. The expression is as follows:

[0093] Y FF1 (t) = NGVal pre (t) - NGVal meas (t),

[0094] where Y FF1 (t) is the bias degree of the fuel valve, NGVal pre (t) is the predicted opening value of the fuel valve, NGVal meas (t) is the opening value of the fuel valve. Calculate the difference between the predicted opening value of the inlet guide vane and the opening value of the inlet guide vane in the collected operation data. The expression is as follows:

[0095] Y FF2 (t) = IGV pre (t) - IGVmeas (t),

[0096] where Y FF2 (t) is the bias of the inlet guide vane, IGV pre (t) is the predicted opening value of the inlet guide vane, IGV meas (t) is the opening value of the inlet guide vane.

[0097] Furthermore, for the bias of the fuel valve, according to the calculation deviation rate of the bias of the fuel valve and the opening value of the fuel valve, its feedforward adjustment signal is determined according to the deviation rate, and its expression is as follows:

[0098]

[0099] where ΔY FF1 (t) is the deviation rate of the fuel valve, Y FF1 (t) is the bias of the fuel valve, Y FF1 (t) = 0.0001, when ΔY FF1 (t) > 0.05, Y FF1 (t) is 0.0001.

[0100] Furthermore, for the bias rate of the inlet guide vane, according to the calculation deviation rate of the bias of the inlet guide vane and the opening value of the inlet guide vane, its feedforward adjustment signal is determined according to the deviation rate, and its expression is as follows:

[0101]

[0102] where ΔY FF2 (t) is the deviation rate of the inlet guide vane, Y FF2 (t) is the bias of the inlet guide vane, Y FF2 (t) = 0.0001, when ΔY FF2 (t) > 0.05, Y FF2 (t) is 0.0001.

[0103] Furthermore, according to the gas turbine exhaust temperature and the predicted exhaust temperature in the real-time operation data, a deviation signal is obtained, and the compensation adjustment signal is obtained according to the deviation signal and the preset signal, and its expression is as follows:

[0104]

[0105] where e(t) is the difference between the gas turbine exhaust temperature and the predicted exhaust temperature, that is, the deviation signal, Y FB (t) is the compensation adjustment signal, and the others are all preset signals, including: K p is the proportional gain, K i is the integral gain, ∫e(τ)dτ is the integral of the deviation, K d is the derivative gain, is the deviation rate.

[0106] In this embodiment, the compensation adjustment signals for the fuel valve and the inlet guide vane are calculated respectively. K p is the proportional gain, K i is the integral gain, ∫e(τ)dτ is the integral of the deviation, K d is the derivative gain, simulates the deviation rate to obtain the compensation adjustment signal Y FB1 (t) of the fuel valve and the compensation adjustment signal Y FB2 (t) of the inlet guide vane.

[0107] Furthermore, according to the feedforward adjustment signal and the compensation adjustment signal of the above fuel valve and inlet guide vane, a weighted fusion is performed to obtain the execution control quantity.

[0108] Specifically, the expressions for obtaining the execution control of the fuel valve and the inlet guide vane are as follows:

[0109] Y1(t) = α1 * Y FF1 (t) + β1 * Y FB1 (t),

[0110] Y2(t) = α2 * Y FF2 (t) + β2 * Y FB2 (t),

[0111] where Y1(t) is the execution control quantity of the fuel valve, Y2(t) is the execution control quantity of the inlet guide vane, Y FB1 (t) is the compensation adjustment signal of the fuel valve, Y FB2 (t) is the compensation adjustment signal of the inlet guide vane, Y FF1 (t) is the bias degree of the fuel valve, Y FF2 (t) is the bias degree of the inlet guide vane, α1 and β1 are the weighting coefficients of the execution control quantity of the fuel valve, α1 + β1 = 1; α2 and β2 are the weighting coefficients of the execution control quantity of the fuel valve, α2 + β2 = 1; in this embodiment, α1 = 0.05 and α2 = 0.1 at the initial input. As the continuously updated deployed training model gradually increases the weight of the feedforward control, that is, increases the α ratio, the range of α1 and α2 is 0.05 - 0.5.

[0112] Furthermore, the real-time operation data is collected through a sliding window to obtain an updated data set, and the adaptable layer of the feedforward prediction model is updated according to the updated data set and the feedforward prediction value through adaptive online gradient and low-rank adaptation.

[0113] In this embodiment, according to the feedforward prediction value of the current model and the real-time operation data, the loss value is calculated, and its expression is as follows:

[0114] L = MSE(y pred , y meas ),

[0115] where L is the loss value, y pred is the feedforward prediction value, and y meas is the real-time operation data. According to the low-rank matrix A and the low-rank matrix B in the above steps, the calculated gradient is expressed by the following formula:

[0116]

[0117] where is the gradient, is the partial derivative of the above loss value L with respect to the low-rank matrix A, is the partial derivative of the above loss value L with respect to the low-rank matrix B.

[0118] Calculate the diagonal estimation of the Fisher matrix according to the gradient. The formula is as follows:

[0119]

[0120] where F t is the diagonal estimation of the Fisher matrix, i.e., the diagonal matrix, and F t-1 is the diagonal estimation of the Fisher matrix at time step t - 1, is the gradient, and γ is the smoothing coefficient, which is selected as 0.9 according to the data characteristics in this embodiment. Update the low-rank matrix A and the low-rank matrix B according to the gradient and the diagonal estimation of the Fisher matrix. The expression is as follows:

[0121]

[0122] where A t+1 is the low-rank matrix A after update, B t+1 is the low-rank matrix B after update, A t is the low-rank matrix A at the current time, and B t is the low-rank matrix B at the current time, is the gradient, and F t is the diagonal estimation of the Fisher matrix.

[0123] Furthermore, update the low-rank matrix A and the low-rank matrix B of the adjustable adaptation layer in the feedforward prediction model to the low-rank matrix A t+1 and the low-rank matrix B t+1 , and complete the online update of the model.

[0124] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0125] Based on another aspect of the embodiments of the present application, the present invention also provides a method and device for controlling the exhaust gas temperature of a hydrogen gas turbine. As Figure 2 shown, the device includes:

[0126] An acquisition module 201, configured to acquire operation data, where the operation data includes historical operation data and real-time operation data, and preprocess the operation data;

[0127] A training module 202, configured to train according to the preprocessed operation data to obtain a feedforward prediction model;

[0128] A control module 203, configured to input the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculate a bias degree according to the feedforward prediction value and the real-time operation data, and obtain a feedforward adjustment signal according to the bias degree; calculate and obtain a compensation adjustment signal through PID control according to the real-time operation data, a preset signal, and the feedforward prediction model; weight the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control amount;

[0129] An update module 204, configured to acquire the real-time operation data through a sliding window to obtain an update data set, and update the adaptable layer of the feedforward prediction model through adaptive online gradient and low rank adaptation according to the update data set and the feedforward prediction value.

[0130] As an optional solution, the above device is further configured to: acquire operation data and preprocess the operation data.

[0131] As an optional solution, the above device is further configured to: preprocess the operation data, perform noise reduction, feature engineering, and standardization on the data; divide it into a training set, a validation set, and a test set.

[0132] As an alternative solution, the above device is further configured to: train based on the preprocessed operation data, construct a Transformer model based on the operation data, use the ambient temperature, the ambient pressure, the ambient humidity, the compressor outlet temperature, the compressor outlet pressure, the gas turbine outlet pressure, the gas turbine power, and the fuel calorific value as input data, and use the opening value of the flow control device and the gas turbine exhaust temperature as output targets to train and obtain the feed-forward prediction model.

[0133] As an alternative solution, the above device is further configured to: the feed-forward prediction value, the predicted opening value of the flow control device, and the predicted exhaust temperature of the gas turbine.

[0134] As an alternative solution, the above device is further configured to: calculate the bias degree based on the feed-forward prediction value and the real-time operation data, and obtain the bias degree by calculating the difference between the opening value of the flow control device in the real-time operation data and the predicted opening value.

[0135] As an alternative solution, the above device is further configured to: obtain the feed-forward adjustment signal based on the bias degree, calculate the deviation rate based on the bias degree and the predicted opening value, and determine the feed-forward adjustment signal based on the deviation rate.

[0136] As an alternative solution, the above device is further configured to: obtain a compensation adjustment signal based on the real-time operation data and a preset signal, obtain a deviation signal based on the gas turbine exhaust temperature and the predicted exhaust temperature in the real-time operation data, and obtain the compensation adjustment signal based on the deviation signal and the preset signal.

[0137] As an alternative solution, the above device is further configured to: update the adaptable layer of the feed-forward prediction model through adaptive online gradient and low-rank adaptation according to the updated data set and the feed-forward prediction value, calculate the loss based on the feed-forward prediction value to obtain a loss value, obtain a gradient based on the loss value, calculate a diagonal matrix based on the gradient, and update the low-rank matrix in the adjustable adaptable layer based on the diagonal matrix and the gradient.

[0138] As an alternative solution, the above device is further configured to: collect the real-time operation data using a sliding window, where the window size is 25 days to 35 days and the window period is 1h to 2h.

[0139] In the embodiments of the present application, the term "module" or "unit" refers to a computer program with a predetermined function or a part of a computer program, which works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Similarly, one processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of an overall module or unit that includes the function of that module or unit.

[0140] Regarding the device in the above embodiments, the specific manner in which each module performs operations has been described in detail in the embodiments related to the method, and will not be elaborated here.

[0141] According to one aspect of the present application, there is provided a computer program product, which includes a computer program.

[0142] The serial numbers of the embodiments of the present application above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0143] Figure 4 A block diagram of a computer system for implementing the electronic device in the embodiments of the present application is schematically shown.

[0144] It should be noted that Figure 4 The computer system 400 of the electronic device shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of the present application.

[0145] As Figure 4 shown, the computer system 400 includes a central processing unit 401 (Central Processing Unit, CPU), which can perform various appropriate actions and processes according to a program stored in a read-only memory 402 (Read-Only Memory, ROM) or a program loaded from a storage section 408 into a random access memory 403 (Random Access Memory, RAM). In the random access memory 403, various programs and data required for system operations are also stored. The central processing unit 401, the read-only memory 402, and the random access memory 403 are connected to each other through a bus 404. An input / output interface 405 (Input / Output interface, i.e., I / O interface) is also connected to the bus 404.

[0146] The following components are connected to the input / output interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a local area network card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output interface 405 as needed. A removable medium 43, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 410 as needed so that a computer program read from it can be installed into the storage section 408 as needed.

[0147] Specifically, according to an embodiment of the present application, the processes described in each method flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 409, and / or installed from the removable medium 43. When the computer program is executed by the central processing unit 401, various functions defined in the system of the present application are executed.

[0148] In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 409, and / or installed from the removable medium 43. When the computer program is executed by the central processing unit 401, various functions provided by the embodiments of the present application are executed.

[0149] According to another aspect of the embodiments of the present application, an electronic device for a method of controlling the exhaust gas temperature of a hydrogen gas turbine is also provided. In this embodiment, the electronic device is taken as an example of a terminal device for illustration. As Figure 5 shown, the electronic device includes a memory 502 and a processor 504. A computer program is stored in the memory 502, and the processor 504 is configured to execute the steps in any of the above method embodiments through the computer program.

[0150] Optionally, in this embodiment, the above-mentioned electronic device can be at least one network device among multiple network devices in a computer network.

[0151] Optionally, in this embodiment, the above-mentioned processor can be configured to execute the methods in the embodiments of the present application through a computer program.

[0152] Optionally, those of ordinary skill in the art can understand, Figure 5The structure shown is only schematic, Figure 5 and it does not limit the structure of the above-mentioned electronic device. For example, the electronic device may further include more or fewer components (such as a network interface, etc.) than those shown Figure 5 in the figure, or have a different configuration from that shown Figure 5 in the figure.

[0153] Among them, the memory 502 can be used to store software programs and modules, such as the program instructions / modules corresponding to a hydrogen gas turbine exhaust temperature control method and device in an embodiment of the present application. The processor 504 executes various functional applications and data processing by running the software programs and modules stored in the memory 502, that is, implements the above-mentioned hydrogen gas turbine exhaust temperature control method. The memory 502 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory 502 may further include a memory remotely set relative to the processor 504, and these remote memories can be connected to the terminal through a network. Examples of the above network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network, and combinations thereof. Among them, the memory 502 can specifically but not limitedly be used to store the collected operation data or cleaned data information. As an example, as Figure 5 shown, the above-mentioned memory 502 may include but are not limited to the data acquisition module 201, training module 202, control module 203, and update module 204 in the above-mentioned hydrogen gas turbine exhaust temperature control device. In addition, it may also include but are not limited to other module units in the above-mentioned device, which will not be elaborated in this example.

[0154] Optionally, the above-mentioned transmission device 506 is used to receive or send data via a network. Specific examples of the above network may include a wired network and a wireless network. In one instance, the transmission device 506 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers through a network cable, thereby enabling communication with the Internet or local area network. In one instance, the transmission device 506 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0155] In addition, the above-mentioned electronic device further includes: a display 508, which is used to display the above-mentioned operation data or cleaned data; and a connection bus 510, which is used to connect each module component in the above-mentioned electronic device.

[0156] In other embodiments, the above-mentioned terminal device or server can be a node in a distributed system. Among them, the distributed system can be a blockchain system, and the blockchain system can be a distributed system formed by connecting the multiple nodes in the form of network communication. Among them, the nodes can form a peer-to-peer network, and any form of computing device, such as electronic devices like servers and terminals, can become a node in the blockchain system by joining the peer-to-peer network.

[0157] According to one aspect of the present application, there is provided a computer-readable storage medium. The processor of the electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device executes a hydrogen gas turbine exhaust temperature control method provided in various optional implementation manners of the above-mentioned hydrogen gas turbine exhaust temperature control.

[0158] Optionally, in this embodiment, the above-mentioned computer-readable storage medium can be set to store for executing the methods in the various embodiments of the present application.

[0159] Optionally, in this embodiment, those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by a program instructing the relevant hardware of the terminal device. The program can be stored in a computer-readable storage medium, and the storage medium can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disc, etc.

[0160] The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0161] If the integrated unit in the above embodiments is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the above-mentioned computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in the storage medium and includes several instructions for causing one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0162] In the above embodiments of the present application, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0163] In several embodiments provided by the present application, it should be understood that the disclosed application programs can be implemented in other ways. Among them, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of units or modules can be in electrical or other forms.

[0164] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0165] In addition, each functional unit in various embodiments of the present application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0166] The above is only the preferred embodiment of the present invention and is not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0167] In summary, from the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects:

[0168] 1. The present invention proposes a method and device for controlling the exhaust gas temperature of a hydrogen gas turbine. Through the feedforward control strategy, it can predict the disturbances in the system in advance, thereby reducing the impact of the disturbances on the system performance, improving the stability and response speed of the system; combining the predicted results with the traditional feedback control to obtain the final execution control amount can more accurately adjust the control amount, realize the optimal control of the system performance, improve the overall operation efficiency and economy of the system; the feedforward control strategy can effectively enhance the robustness of the system, so that it can still maintain good performance and stability in the face of complex operating environments and uncertain factors, and reduce the probability of faults; the compensation adjustment strategy can effectively eliminate the steady-state error through the integral part of the traditional PID controller, ensuring that the system maintains high precision during long-term operation.

[0169] 2. The present invention provides a method and device for controlling the exhaust temperature of a hydrogen gas turbine. By introducing an online learning and model updating mechanism, it can monitor the operating state of the unit in real time and update the model according to the latest operating data, thus ensuring that the model can always accurately reflect the actual operating conditions of the unit, improving the adaptability and optimization ability of the system; the online learning and model updating mechanism continuously obtains real-time operating data to achieve continuous improvement and optimization of the system performance. Through continuous learning and adjustment, the system can better cope with various changes and challenges, improve the overall operating efficiency and economy of the system, and at the same time can promptly detect and solve potential problems, improve the reliability and stability of the system, reduce the probability of failures, and ensure the long-term stable operation of the system.

[0170] 3. The present invention provides a method and device for controlling the exhaust temperature of a hydrogen gas turbine. By applying an advanced Transformer model, it can accurately predict the exhaust temperature of the hydrogen gas turbine, thereby achieving precise control and optimization of the exhaust temperature. Utilizing the powerful modeling ability of the model, it can comprehensively consider various influencing factors, such as fuel flow rate, air flow rate, ambient temperature, etc., so as to achieve comprehensive optimization of the performance of the gas turbine, improve the overall operating efficiency and stability; by real-time monitoring the exhaust temperature and combining with the feedforward prediction value of the feedforward prediction model, it can promptly detect abnormal situations and make adjustments to ensure that the gas turbine operates in the best state and extend the service life of the equipment.

[0171] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and modifications can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0172] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0173] It should be noted that in the description of this specification, the descriptions referring to the reference terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. mean 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 invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

Claims

1. A method for controlling the exhaust gas temperature of a hydrogen gas turbine, characterized in that Including: Collecting operation data, where the operation data includes historical operation data and real-time operation data; Training based on the preprocessed operation data to obtain a feedforward prediction model, where the feedforward prediction model includes an adjustable adaptation layer; Inputting the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculating a bias degree based on the feedforward prediction value and the real-time operation data, and obtaining a feedforward adjustment signal based on the bias degree; Obtaining a compensation adjustment signal based on the real-time operation data and a preset signal; Weighting the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control quantity.

2. The method according to claim 1, characterized in that, Also including: Collecting the real-time operation data through a sliding window to obtain an updated data set, and updating the adaptable layer of the feedforward prediction model through adaptive online gradient and low-rank adaptation based on the updated data set and the feedforward prediction value.

3. The method according to claim 1, wherein The operation data includes: Compressor outlet pressure, gas turbine outlet pressure, opening value of the flow control device, gas turbine power, fuel calorific value, and gas turbine exhaust temperature.

4. The method according to claim 3, characterized in that The operation data also includes: Ambient temperature, ambient pressure, ambient humidity, and / or compressor outlet temperature.

5. The method according to claim 3, wherein The flow control device includes: Fuel valve and / or inlet guide vane.

6. The method according to claim 1, wherein Collecting operation data also includes: Preprocessing the operation data.

7. The method according to claim 6, wherein Preprocessing the operation data includes: Denosing, feature engineering, and standardizing the data; Dividing it into a training set, a validation set, and a test set.

8. The method according to claim 4, wherein Training based on the preprocessed operation data includes: Based on the operation data, constructing a Transformer model, using the ambient temperature, the ambient pressure, the ambient humidity, the compressor outlet temperature, the compressor outlet pressure, the gas turbine outlet pressure, the gas turbine power, and the fuel calorific value as input data, and using the opening value of the flow control device and the gas turbine exhaust temperature as output targets, training to obtain the feedforward prediction model.

9. The method according to claim 3, wherein The feedforward prediction value includes: The predicted opening value of the flow control device and the predicted exhaust temperature of the gas turbine.

10. The method according to claim 9, wherein Calculating the bias degree based on the feedforward prediction value and the real-time operation data includes: Obtaining the bias degree by calculating the difference between the opening value of the flow control device in the real-time operation data and the predicted opening value.

11. The method according to claim 10, wherein Obtaining the feedforward adjustment signal based on the bias degree includes: Calculating a deviation rate based on the bias degree and the predicted opening value, and determining the feedforward adjustment signal based on the deviation rate.

12. The method according to claim 9, wherein Obtaining the compensation adjustment signal based on the real-time operation data and the preset signal includes: Obtaining a deviation signal based on the gas turbine exhaust temperature and the predicted exhaust temperature in the real-time operation data, and obtaining the compensation adjustment signal based on the deviation signal and the preset signal.

13. The method according to claim 2, wherein Updating the adaptable layer of the feedforward prediction model through adaptive online gradient and low-rank adaptation based on the updated data set and the feedforward prediction value includes: Calculate the loss according to the feedforward prediction value to obtain a loss value, obtain the gradient according to the loss value, calculate a diagonal matrix according to the gradient, and update the low-rank matrix in the adjustable adaptation layer according to the diagonal matrix and the gradient.

14. The method according to claim 2, wherein Collect the real-time operation data by using a sliding window, including: The window size is 25 days - 35 days, and the window period is 1h - 2h.

15. A hydrogen gas turbine exhaust gas temperature control device, characterized in that, Including: A collection module for collecting operation data, where the operation data includes historical operation data and real-time operation data, and preprocessing the operation data; A training module for training according to the preprocessed operation data to obtain a feedforward prediction model; A control module for inputting the real-time operation data into the feedforward prediction model to obtain a feedforward prediction value, calculating a bias degree according to the feedforward prediction value and the real-time operation data, and obtaining a feedforward adjustment signal according to the bias degree; Calculate and obtain a compensation adjustment signal through PID control according to the real-time operation data and a preset signal; weight the feedforward adjustment signal and the compensation adjustment signal to obtain an execution control amount; An update module for collecting the real-time operation data through a sliding window to obtain an updated data set, and updating the adaptable layer of the feedforward prediction model according to the updated data set and the feedforward prediction value through adaptive online gradient and low-rank adaptation.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein the computer program, when executed by an electronic device, executes the method described in any one of claims 1 to 14.

17. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 14 are implemented.

18. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to execute the method described in any one of claims 1 to 14 through the computer program.

Images