Turbine outlet temperature adaptive limiting protection method, device, equipment and medium

Abstract

This application discloses a method, device, equipment, and medium for adaptive limiting protection of turbine outlet temperature. The method includes the following steps: S1, acquiring the state parameters of a turboshaft engine and calculating the corresponding eigenvalue matrix; S2, the onboard model of the turboshaft engine predicts the engine performance parameters online based on the state parameters and inputs them into an unscented Kalman state estimator. The onboard model of the turboshaft engine includes an input-feedback dual adaptive long short-term memory network; S3, the unscented Kalman state estimator estimates the gas turbine outlet temperature online based on the predicted engine performance parameters; S4, when the estimated gas turbine outlet temperature exceeds the temperature limit, the fuel flow rate of the engine is corrected online to achieve adaptive limiting protection control of the gas turbine outlet temperature. This application reduces the measurement error of the gas turbine outlet thermocouple temperature sensor and the probability of misjudging and forgetting some strongly correlated input parameters caused by uncertain time delay abrupt changes, thereby improving control accuracy and real-time performance.

Description

Technical Field

[0001] This application relates to the field of aero-engine technology, and in particular to a method, apparatus, device, and medium for adaptive limiting protection of turbine outlet temperature. Background Technology

[0002] In recent years, driven by both the accelerated modernization of national defense and the growing demand for civil aviation, aero-engine technology has experienced leapfrog development. However, with the continuous improvement of high-performance requirements, aero-engines are operating at their performance limits, and performance improvements often involve pushing beyond existing safety boundaries. Performance and safety are mutually restrictive; pursuing extreme performance can easily lead to engine safety issues, making over-limit safety protection crucial.

[0003] For aero-engine control systems, the goal is to ensure stable and reliable engine operation under any environment and operating condition, maximizing performance benefits. Besides controlled variables, aero-engine control systems contain many limiting variables, among which the most critical are the low-pressure rotor's physical speed, compressor outlet static pressure, and turboshaft engine gas turbine outlet temperature. In certain situations, these physical quantities may exceed their maximum limits, leading to serious consequences. For example, exceeding the limit for turboshaft engine gas turbine outlet temperature can cause turbine blade ablation, severely impacting turbine blade lifespan and threatening engine operational safety. Currently, turboshaft engine gas turbine outlet temperature is often controlled by incorporating a limiting protection control module based on sensor-measured temperature signals into the control structure. However, due to the inherent thermal inertia of thermocouple sensors, the steady-state error of temperature measurements is large, and dynamic characteristics exhibit a certain lag. Protection against turbine outlet temperature is often based on compensation feedback, at which point the turbine is already in an over-temperature state, failing to provide early protection and exhibiting a time lag. Summary of the Invention

[0004] This application provides a turbine outlet temperature adaptive limiting protection method to solve the technical problems of existing turboshaft engine gas turbine outlet temperature protection methods, which do not provide early protection for the turbine and have large errors and time lag.

[0005] This application is achieved through the following solution:

[0006] The turbine outlet temperature adaptive limiting protection method includes the following steps:

[0007] S1. Obtain the state parameters of the turboshaft engine and calculate the corresponding eigenvalue matrix. The eigenvalues ​​are used to characterize the correlation between the state parameters and the output, and the magnitude of the eigenvalues ​​is proportional to the correlation.

[0008] S2. The airborne model of the turboshaft engine predicts the engine performance parameters online based on the state parameters of the turboshaft engine and inputs them into the unscented Kalman state estimator. The airborne model of the turboshaft engine includes an input-feedback dual adaptive long short-term memory network. The input-feedback dual adaptive long short-term memory network adaptively allocates the weights of the input gates according to the size of the eigenvalues, and adaptively corrects the feedback weights corresponding to each input parameter in the forget gate according to the size of the eigenvalues.

[0009] S3. The unscented Kalman state estimator estimates the gas turbine outlet temperature T online based on the predicted engine performance parameters. 45 ;

[0010] S4, when the online estimated gas turbine outlet temperature T 45 When the temperature limit is exceeded, the temperature limit is used as a reference command and the temperature estimate is used as the controlled parameter. The control quantity is calculated using a PI controller to correct the fuel flow of the engine online and realize adaptive limit protection control of the gas turbine outlet temperature.

[0011] Further, step S1 specifically includes the following steps:

[0012] S11. Obtain the status parameters of the turboshaft engine, including flight altitude, flight speed, gas turbine speed, power turbine speed, collective pitch, fuel flow rate, and exhaust volume.

[0013] S12. Normalize the acquired turboshaft engine state parameters.

[0014] S13. Fit the normalized state parameters of the turboshaft engine to obtain the mapping matrix between input and output, and obtain the eigenvalue matrix through singular value decomposition.

[0015] Furthermore, step S12 specifically includes the following steps:

[0016] The input and output sample sequences are normalized based on the mathematical expectation and standard deviation of the signal to unify the evaluation criteria.

[0017] ;

[0018] In the formula, X Indicates input data ，X out This represents the normalized input data. μ and σ yes X Expected value and standard deviation ε These are small positive numbers added for the stability of division.

[0019] Furthermore, S13 specifically includes the following steps:

[0020] S131. Obtain input data through least squares fitting. X To output data Y Mapping matrix:

[0021] ;

[0022] S132. Obtain the eigenvalues ​​by decomposing the above equation using singular values:

[0023] ;

[0024] In the formula, U M express M orthogonal matrix of order 1, V N express N orthogonal matrix of order, r =rank( C M ), representing the mapping matrix C m rank, κ 1. κ 2…、 κ r for r Each feature value.

[0025] Furthermore, in step S2, the specific strategy for the input-feedback dual adaptive long short-term memory network to adaptively allocate the weights of the input gates based on the feature value is as follows:

[0026] ;

[0027] In the formula, tanh(•) is the hyperbolic tangent function. W ij The first input gate in the Long Short-Term Memory network j The initial weight values ​​corresponding to each input parameter. For the input gate, the first j The weights of the input parameters are adaptively assigned;

[0028] The input-feedback dual adaptive long short-term memory network adaptively adjusts the feedback weights corresponding to each input parameter in the forget gate based on the feature value. The specific strategy is as follows:

[0029] ;

[0030] In the formula, W fj The first of the forgetting gates in the Long Short-Term Memory Network j The initial feedback weight values ​​corresponding to each input. This is the correction value for the corresponding feedback weight value.

[0031] Further, in step S2, the forward propagation process of the input-feedback dual adaptive long short-term memory network is as follows:

[0032] ;

[0033] In the formula, f t , i t , o t These represent the activation values ​​of the forget gate, input gate, and output gate, respectively. , C t This indicates the candidate cell state and the current cell state; h t Indicates output in hidden state; σ Represents the sigmoid activation function; ⊙ represents the Hadamard product. W i , W f , W C These are the input, feedback, and cell weight matrices, respectively.

[0034] Further, in step S2, the airborne model of the turboshaft engine is:

[0035] ;

[0036] Among them, the airborne model of the turboshaft engine is based on the current k Moment and the two previous moments in history k -1、 k fuel flow at time -2 W f Collective rotor pitch θ 0 Intermediate stage exhaust volume of the compressor L ct Gas turbine speeds at the two most recent historical moments N g Power turbine speed N p Gas turbine outlet temperature T 45 Flight altitude H and forward flight speed V c For input; current k Gas turbine speed at any given time N g Power turbine speed N p Engine output torqueT qe For output, f UAFLSTM (•) is the input and output mapping function of the airborne model of the turboshaft engine based on the input-feedback dual adaptive long short-term memory network.

[0037] This application also provides a turbine outlet temperature adaptive limiting protection device, including:

[0038] The eigenvalue calculation module is used to obtain the state parameters of the turboshaft engine and calculate the corresponding eigenvalue matrix. The eigenvalues ​​are used to characterize the correlation between the state parameters and the output, and the magnitude of the eigenvalues ​​is proportional to the correlation.

[0039] An engine performance prediction module is used to predict engine performance parameters online based on the turboshaft engine state parameters of the airborne model and input them into an unscented Kalman state estimator. The airborne model of the turboshaft engine includes an input-feedback dual adaptive long short-term memory network. The input-feedback dual adaptive long short-term memory network adaptively allocates the weights of the input gates according to the size of the feature values, and adaptively corrects the feedback weights corresponding to each input parameter in the forget gate according to the size of the feature values.

[0040] A gas turbine outlet temperature estimation module is used by the unscented Kalman state estimator to estimate the gas turbine outlet temperature T online based on predicted engine performance parameters. 45 ;

[0041] The fuel flow control module is used to control the gas turbine outlet temperature T online. 45 When the temperature limit is exceeded, the temperature limit is used as a reference command and the temperature estimate is used as the controlled parameter. The control quantity is calculated using a PI controller to correct the fuel flow of the engine online and realize adaptive limit protection control of the gas turbine outlet temperature.

[0042] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the turbine outlet temperature adaptive limiting protection method.

[0043] This application also provides a storage medium including a stored program that, when the program is executed, controls the device containing the storage medium to perform the steps of the turbine outlet temperature adaptive limiting protection method.

[0044] Compared with the prior art, this application has the following advantages:

[0045] This application proposes a turbine outlet temperature adaptive limiting protection method, device, equipment, and medium. The turbine outlet temperature adaptive limiting protection method introduces an input-feedback dual adaptive long short-term memory network into the airborne model of a turboshaft engine. This long short-term memory network uses dual adaptive correction by setting feedback weights based on eigenvalues ​​for adaptive correction and feedback weights for adaptive correction. Under different operating conditions, the long short-term memory network can automatically analyze and identify input features strongly correlated with the target output for different input combinations (fuel, guide vanes, intermediate stage venting) of the turboshaft engine. Combined with the weight adaptive correction strategy, it can more quickly and accurately construct the mapping relationship between input and output. The model can dynamically adjust the feedback intensity according to the importance (eigenvalue magnitude) of the input features, thereby adaptively correcting the feedback weights corresponding to each input parameter in the forget gate. It has high dynamic accuracy, thereby reducing the probability of some strongly correlated input parameters being misjudged and forgotten due to measurement errors of the gas turbine outlet thermocouple temperature sensor and uncertain time delay abrupt changes. In summary, this application can automatically analyze and identify input features that are strongly correlated with the target output, and combine the weight adaptive correction strategy of the input and feedback matrices to construct a mapping model between input and output more quickly and accurately, with high dynamic accuracy.

[0046] In addition to the purposes, features, and advantages described above, this application has other purposes, features, and advantages. A further detailed description of this application will be provided below with reference to the figures. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0049] Figure 1 This is a flowchart illustrating the turbine outlet temperature adaptive limiting protection method according to a preferred embodiment of this application.

[0050] Figure 2 This is a schematic diagram of a long short-term memory network structure based on input-feedback dual adaptive structure according to a preferred embodiment of this application;

[0051] Figure 3 This is a schematic diagram illustrating the principle of the turbine outlet temperature adaptive limiting protection method according to a preferred embodiment of this application;

[0052] Figure 4 This is a schematic diagram illustrating the relative test error of the gas turbine speed after offline training of the engine in this application;

[0053] Figure 5 This is a schematic diagram illustrating the relative test error of the power turbine speed after offline training of the engine in this application;

[0054] Figure 6 This is a schematic diagram illustrating the relative test error of the engine output torque after offline training of the engine in this application;

[0055] Figure 7 This is a schematic diagram comparing the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 2000m and flight speeds of 10m / s, as presented in this application.

[0056] Figure 8 This is a schematic diagram showing the error distribution between the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 2000m and flight speeds of 10m / s, as presented in this application.

[0057] Figure 9 This is a schematic diagram comparing the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 1000m and flight speeds of 10m / s, as presented in this application.

[0058] Figure 10 This is a schematic diagram showing the error distribution between the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 1000m and flight speeds of 10m / s, as presented in this application.

[0059] Figure 11 This is a schematic diagram comparing the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 500m and flight speeds of 10m / s, as presented in this application.

[0060] Figure 12 This is a schematic diagram showing the error distribution between the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 500m and flight speeds of 10m / s, as presented in this application.

[0061] Figure 13 This is a schematic diagram comparing the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 1500m and flight speeds of 20m / s, as presented in this application.

[0062] Figure 14 This is a schematic diagram showing the error distribution between the estimated and actual values ​​of the gas turbine outlet temperature at flight altitudes of 1500m and flight speeds of 20m / s, as presented in this application.

[0063] Figure 15 This is a schematic diagram illustrating the changes in various parameters before and after the protection is limited under different flight conditions in this application;

[0064] Figure 16 This is a schematic diagram of the turbine outlet temperature adaptive limiting protection device module according to a preferred embodiment of this application;

[0065] Figure 17 This is a schematic block diagram of an electronic device according to a preferred embodiment of this application;

[0066] Figure 18 This is an internal structural diagram of a computer device according to a preferred embodiment of this application. Detailed Implementation

[0067] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0068] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0069] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a turbine outlet temperature adaptive limiting protection device capable of achieving the above functions. The following description uses a turbine outlet temperature adaptive limiting protection device as the executing entity to illustrate this embodiment and the subsequent embodiments.

[0070] like Figure 1 As shown, a preferred embodiment of this application provides a turbine outlet temperature adaptive limiting protection method, including the following steps:

[0071] S1. Obtain the state parameters of the turboshaft engine and calculate the corresponding eigenvalue matrix. The eigenvalues ​​are used to characterize the correlation between the state parameters and the output, and the magnitude of the eigenvalues ​​is proportional to the correlation.

[0072] S2. The airborne model of the turboshaft engine predicts engine performance parameters online based on the turboshaft engine's state parameters and inputs them into an unscented Kalman state estimator. The airborne model of the turboshaft engine includes an input-feedback dual adaptive long short-term memory network (see...). Figure 2 The input-feedback dual adaptive long short-term memory network adaptively allocates the weights of the input gates according to the size of the feature values, and adaptively corrects the feedback weights corresponding to each input parameter in the forget gate according to the size of the feature values.

[0073] S3. The unscented Kalman state estimator estimates the gas turbine outlet temperature T online based on the predicted engine performance parameters. 45 ;

[0074] S4, when the online estimated gas turbine outlet temperature T 45When the temperature limit is exceeded, the temperature limit is used as a reference command and the temperature estimate is used as the controlled parameter. The control quantity is calculated using a PI controller to correct the fuel flow of the engine online and realize adaptive limit protection control of the gas turbine outlet temperature.

[0075] Based on this, this embodiment rapidly establishes a high-precision airborne model of the turboshaft engine for online prediction of engine performance parameters, and utilizes an unscented Kalman state estimator to estimate the gas turbine outlet temperature T. 45 Online estimation. Finally, based on the closed-loop feedback controller, when the gas turbine outlet temperature exceeds the limit, the temperature limit value is used as the reference command, and the temperature estimate value is used as the controlled parameter. The control quantity is calculated using a PI controller to correct the engine's fuel flow online; otherwise, no correction is made. This achieves adaptive limit protection control of the gas turbine outlet temperature. The control principle is as follows: Figure 3 As shown. Additionally, the power turbine speed command, as an external setpoint, sets requirements for the operating conditions of the engine's power turbine. Both the internal and external loop controllers use PI control algorithms. The external loop controller adjusts the power turbine speed to maintain the power turbine speed command; the internal loop controller adjusts the gas turbine speed and can improve the disturbance rejection performance of the external loop controller. In actual operation, the actual speeds of both the gas turbine and power turbine can be collected by their respective sensors.

[0076] This embodiment proposes an adaptive limiting protection method for turbine outlet temperature. This method introduces an input-feedback dual adaptive long short-term memory (LSTM) network into the airborne model of a turboshaft engine. This LSM network employs dual adaptive correction through adaptive correction of feedback weights based on eigenvalues ​​and adaptive correction of feedback weights. This allows the LSM network to automatically analyze and identify input features strongly correlated with the target output under different operating conditions and for different input combinations of the turboshaft engine (fuel, guide vanes, intermediate stage bleed). Combined with a weight adaptive correction strategy, it can more quickly and accurately construct the mapping relationship between input and output. The model can dynamically adjust the feedback intensity according to the importance (eigenvalue magnitude) of the input features, thereby adaptively correcting the feedback weights corresponding to each input parameter in the forget gate. This results in high dynamic accuracy and reduces the probability of misjudging and forgetting some strongly correlated input parameters caused by measurement errors of the gas turbine outlet thermocouple temperature sensor and uncertain time delay abrupt changes. In summary, this embodiment can automatically analyze and identify input features strongly correlated with the target output, and, combined with a weight adaptive correction strategy for the input and feedback matrices, more quickly and accurately construct a mapping model between input and output with high dynamic accuracy.

[0077] Preferably, the unscented Kalman state estimator can be replaced with other types of filters, such as the extended Kalman state estimator.

[0078] Preferably, step S1 specifically includes the following steps:

[0079] S11. Obtain the status parameters of the turboshaft engine, including flight altitude, flight speed, gas turbine speed, power turbine speed, collective pitch, fuel flow rate, and exhaust volume.

[0080] S12. Normalize the acquired turboshaft engine state parameters.

[0081] S13. Fit the normalized state parameters of the turboshaft engine to obtain the mapping matrix between input and output, and obtain the eigenvalue matrix through singular value decomposition.

[0082] Preferably, step S12 specifically includes the following steps:

[0083] The input and output sample sequences are normalized based on the mathematical expectation and standard deviation of the signal to unify the evaluation criteria.

[0084] ;

[0085] In the formula, X Indicates input data ，X out This represents the normalized input data. μ and σ yes X Expected value and standard deviation ε These are small positive numbers added for the stability of division.

[0086] Preferably, S13 specifically includes the following steps:

[0087] S131. Obtain input data through least squares fitting. X To output data Y Mapping matrix:

[0088] ;

[0089] S132. Obtain the eigenvalues ​​by decomposing the above equation using singular values:

[0090] ;

[0091] In the formula, U M express M orthogonal matrix of order 1, V N express N orthogonal matrix of order 1, r =rank( C M ), representing the mapping matrixC m rank, κ 1. κ 2…、 κ r for r Each feature value.

[0092] Considering that during the operation of a turboshaft engine, the guide vane angle, the compressor intermediate stage bleed gas volume, and the timing of fuel flow are coordinated to regulate the engine, such as opening the compressor intermediate stage bleed gas valve at low speeds and closing the bleed gas valve at high speeds, different input combinations (fuel, fuel-guide vane, fuel-guide vane-bleed gas, fuel-bleed gas, etc.) exist throughout the engine's operation. Furthermore, the correlation or relationship between inputs and outputs varies significantly under different input combinations. To address this, this application introduces an adaptive input weight allocation and adaptive feedback weight correction mechanism, specifically including:

[0093] In step S2, the specific strategy for the input-feedback dual adaptive long short-term memory network to adaptively allocate the weights of the input gates based on the feature values ​​is as follows:

[0094] ;

[0095] In the formula, tanh(•) is the hyperbolic tangent function. W ij The first input gate in the Long Short-Term Memory network j The initial weight values ​​corresponding to each input parameter. For the input gate, the first j The weights of the input parameters are adaptively assigned;

[0096] The input-feedback dual adaptive long short-term memory network adaptively adjusts the feedback weights corresponding to each input parameter in the forget gate based on the feature value. The specific strategy is as follows:

[0097] ;

[0098] In the formula, W fj The first of the forgetting gates in the Long Short-Term Memory Network j The initial feedback weight values ​​corresponding to each input. This is the correction value for the corresponding feedback weight value.

[0099] Eigenvalues ​​can characterize the correlation between an input and an output to a certain extent; the larger the eigenvalue, the stronger the correlation, and vice versa. Based on this, state parameters with large eigenvalues ​​are preferred as inputs, and the weights of the input gates are adjusted using these eigenvalues. Input parameters with stronger correlation receive larger weights, and vice versa. Larger weights for highly correlated parameters act as a "reward" mechanism, leading to more accurate predictions. Furthermore, considering the measurement errors and uncertain time delays of the gas turbine outlet thermocouple temperature sensor, this embodiment introduces an adaptive feedback weight correction function. This allows the recording model to dynamically adjust the feedback strength based on the importance (eigenvalue) of the input features, thereby adaptively correcting the feedback weights corresponding to each input parameter in the forget gate.

[0100] In this embodiment, the feedback weights corresponding to each input parameter in the forget gate are adaptively adjusted based on the magnitude of the eigenvalues. The larger the eigenvalue, the more significant the enhancement of the feedback weights. The forget gate of the Long Short-Term Memory network determines which information from the previous time step needs to be retained or forgotten through σ (the sigmoid activation function, with an output value between [0,1]). By introducing the aforementioned adaptive feedback weight adjustment module, the feedback weights are adaptively strengthened based on the correlation of the input features, reducing the probability of some strongly correlated input parameters being misjudged and forgotten due to sensor measurement errors and uncertain time delay abrupt changes.

[0101] Based on the above formula, in step S2, the forward propagation process of the input-feedback dual adaptive long short-term memory network is as follows:

[0102] ;

[0103] In the formula, f t , i t , o t These represent the activation values ​​of the forget gate, input gate, and output gate, respectively. , C t This indicates the candidate cell state and the current cell state; h t Indicates output in hidden state; σ Represents the sigmoid activation function; ⊙ represents the Hadamard product. W i , W f , W C These are the input, feedback, and cell weight matrices, respectively.

[0104] The introduction of the input-feedback weight dual adaptive module enables the long short-term memory network to automatically analyze and identify input features that are strongly correlated with the target output under different operating conditions and for different input combinations (fuel, guide vanes, intermediate stage venting) of the turboshaft engine. Combined with the aforementioned weight adaptive correction strategy, it can more quickly and accurately construct the mapping relationship between input and output, and has high dynamic accuracy.

[0105] Preferably, in step S2, the airborne model of the turboshaft engine is:

[0106] ;

[0107] Among them, the airborne model of the turboshaft engine is based on the current k Moment and the two previous moments in history k -1、 k fuel flow at time -2 W f Collective rotor pitch θ 0 Intermediate stage exhaust volume of the compressor L ct Gas turbine speeds at the two most recent historical moments N g Power turbine speed N p Gas turbine outlet temperature T 45 Flight altitude H and forward flight speed V c For input; current k Gas turbine speed at any given time N g Power turbine speed N p Engine output torque T qe For output, f UAFLSTM (•) is the input and output mapping function of the turboshaft engine airborne model established based on the input-feedback dual adaptive long short-term memory network. Since the turboshaft engine can be approximated as a second-order system, the above-mentioned turboshaft engine airborne model can be constructed.

[0108] Selecting the relative speed of the power turbine N p relative speed of gas turbine N g Engine output torque T qeAs state variables in the component-level model of the turboshaft engine, the input to the unscented Kalman state estimator is the error between the gas turbine speed output from the airborne model and the measured speed, and the output is the gas turbine outlet temperature T. 45 The unscented Kalman state estimator is based on the engine k The state variables and inputs at time -1 can be estimated. k The state variable at any given time.

[0109] Based on a component-level model used to simulate the operating characteristics of a real turboshaft engine, input and output sample data for training the airborne model of the turboshaft engine were collected at different flight envelope points. The relative test error is as follows: Figures 4 to 6 As shown. Figure 4 and Figure 5 As shown, the gas turbine speed N g With power turbine speed N p The relative error does not exceed 1%, such as Figure 6 As shown, the engine output torque T qe The relative error is basically no more than 2%, confirming that the established airborne model has high accuracy. The unscented Kalman filter (UKF) estimates the state variables based on the residual vector between the output parameters of the engine and the airborne nonlinear model. It is an observation algorithm that directly applies the principle of unscented transformation to the integral calculation in Bayes recursive estimation.

[0110] Based on the component-level model and airborne model of the turboshaft engine, the accuracy of the established unscented Kalman state estimator was verified under different flight conditions (flight altitude and flight speed). The results are as follows: Figures 7 to 14 As shown. Under different operating conditions, T based on the unscented Kalman state estimator 45 The estimated value has an error of less than 1.4% compared with the output value of the turboshaft engine component-level model, indicating high estimation accuracy.

[0111] When the estimated gas turbine outlet temperature exceeds the limit, to achieve the purpose of limiting protection, the limit is used as a reference command, and the estimated gas turbine outlet temperature is used as the controlled variable. This allows the PI controller to generate a virtual control variable and correct the engine's fuel flow online. The structure diagram of this limiting protection method is shown below. Figure 1 As shown.

[0112] Figure 15 The results of over-limit protection control under two flight conditions are presented. In state A, overheating occurs at t=35s. Under the action of the limiting protection controller, the fuel flow rate W... fb Decrease, N p The sag also increases accordingly, and the gas turbine outlet temperature T 45The design temperature was reduced from 118% to the limit of 110%, and the over-temperature time was shortened by more than 50%; in state B, under the action of the limit protection controller, T 45 The design temperature at this point is reduced from 123% to 110%, which demonstrates the superiority and robustness of the turbine outlet temperature adaptive limiting protection method of this application.

[0113] like Figure 16 As shown, another preferred embodiment of this application also provides a turbine outlet temperature adaptive limiting protection device, including:

[0114] The eigenvalue calculation module is used to obtain the state parameters of the turboshaft engine and calculate the corresponding eigenvalue matrix. The eigenvalues ​​are used to characterize the correlation between the state parameters and the output, and the magnitude of the eigenvalues ​​is proportional to the correlation.

[0115] An engine performance prediction module is used to predict engine performance parameters online based on the turboshaft engine state parameters of the airborne model and input them into an unscented Kalman state estimator. The airborne model of the turboshaft engine includes an input-feedback dual adaptive long short-term memory network. The input-feedback dual adaptive long short-term memory network adaptively allocates the weights of the input gates according to the size of the feature values, and adaptively corrects the feedback weights corresponding to each input parameter in the forget gate according to the size of the feature values.

[0116] A gas turbine outlet temperature estimation module is used by the unscented Kalman state estimator to estimate the gas turbine outlet temperature T online based on predicted engine performance parameters. 45 ;

[0117] The fuel flow control module is used to control the gas turbine outlet temperature T online. 45 When the temperature limit is exceeded, the temperature limit is used as a reference command and the temperature estimate is used as the controlled parameter. The control quantity is calculated using a PI controller to correct the fuel flow of the engine online and realize adaptive limit protection control of the gas turbine outlet temperature.

[0118] The turbine outlet temperature adaptive limiting protection device provided in this embodiment adopts the turbine outlet temperature adaptive limiting protection method in the above embodiments, solving the technical problems of existing turboshaft engine gas turbine outlet temperature protection methods that do not provide early protection for the turbine, resulting in large errors and time lags. Compared with the prior art, the beneficial effects of the turbine outlet temperature adaptive limiting protection device provided in this application are the same as those of the turbine outlet temperature adaptive limiting protection method provided in the above embodiments, and other technical features in the turbine outlet temperature adaptive limiting protection device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0119] like Figure 17As shown, a preferred embodiment of this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the turbine outlet temperature adaptive limiting protection method in the above embodiments.

[0120] This application provides an electronic device that employs the turbine outlet temperature adaptive limiting protection method described in the above embodiments. This addresses the technical problems of existing turboshaft engine gas turbine outlet temperature protection methods, which lack advance turbine protection, resulting in large errors and time lags. Compared to the prior art, the electronic device provided in this application has the same beneficial effects as the turbine outlet temperature adaptive limiting protection method provided in the above embodiments, and other technical features of the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0121] like Figure 18 As shown, a preferred embodiment of this application also provides a computer device, which may be a terminal or a liveness detection server, and its internal structure diagram may be as follows. Figure 18 As shown. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with other external computer devices via a network connection. When the computer program is executed by the processor, it implements the steps of the aforementioned turbine outlet temperature adaptive limiting protection method.

[0122] Those skilled in the art will understand that Figure 18 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0123] The computer device provided in this application employs the turbine outlet temperature adaptive limiting protection method described in the above embodiments, solving the technical problems of existing turboshaft engine gas turbine outlet temperature protection methods that do not provide advance protection for the turbine, resulting in large errors and time lags. Compared with the prior art, the beneficial effects of the computer device provided in this application are the same as those of the turbine outlet temperature adaptive limiting protection method provided in the above embodiments, and other technical features in the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0124] A preferred embodiment of this application also provides a storage medium including a stored program that, when the program is executed, controls the device containing the storage medium to perform the steps of the turbine outlet temperature adaptive limiting protection method in the above embodiments.

[0125] It is evident that excessively high turbine outlet temperatures in turboshaft engines severely impact turbine blade lifespan, posing a threat to engine operational safety. Therefore, existing technologies employ a temperature-sensing sensor-based limit protection control module within the control structure. However, the inherent thermal inertia of thermocouple sensors results in significant steady-state errors and dynamic hysteresis in temperature measurements. Therefore, rapidly and accurately sensing transient changes in turbine outlet temperature and applying them to fuel control would significantly contribute to the safe, stable, and reliable operation of the engine.

[0126] Therefore, this application provides an airborne model of a temporal convolutional attention mechanism network, combined with a state estimator based on an unscented Kalman filter, to estimate the gas turbine outlet temperature online. When the gas turbine outlet temperature exceeds the limit, the temperature limit is used as a reference command, and the temperature estimate is used as the controlled parameter. A PI controller is used to calculate the control quantity to correct the engine's fuel flow online; otherwise, no correction is made, achieving the purpose of adaptive limit protection for the gas turbine outlet temperature. The same principle can also be used to achieve speed limit protection.

[0127] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0128] If the functions described in this embodiment are implemented as software functional units and sold or used as independent products, they can be stored in one or more computing device-readable storage media. Based on this understanding, the parts of this application's embodiments that contribute to the prior art or the technical solutions can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computing device (which may be a personal computer, server, mobile computing device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0129] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language C++ and the embedded programming language C.

[0130] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0131] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0132] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0133] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the turbine outlet temperature adaptive limiting protection method as described above.

[0134] The computer program product provided in this application solves the technical problem that existing turboshaft engine gas turbine outlet temperature protection methods do not provide advance protection for the turbine, resulting in large errors and time lags. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the turbine outlet temperature adaptive limiting protection method provided in the above embodiments, and will not be repeated here.

[0135] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0136] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A turbine outlet temperature adaptive limit protection method, characterized by, The method comprises the steps of: S1, obtaining the state parameters of the turboshaft engine and calculating the corresponding eigenvalue matrix, wherein the eigenvalue is used to represent the correlation between the state parameters and the output, and the size of the eigenvalue is proportional to the correlation; S2, the on-line prediction of the engine performance parameters by the turboshaft engine on-board model according to the state parameters of the turboshaft engine and inputting to the unscented Kalman state estimator, wherein the turboshaft engine on-board model comprises an input-feedback double adaptive long short-term memory network, wherein the input-feedback double adaptive long short-term memory network adaptively allocates the weight of the input gate according to the size of the eigenvalue, and adaptively corrects the feedback weight corresponding to each input parameter in the forgetting gate according to the size of the eigenvalue; S3. The unscented Kalman state estimator estimates the gas turbine outlet temperature T online from the predicted engine performance parameters 45 ; S4, when the online estimated gas turbine outlet temperature T 45 When the temperature limit value is exceeded, the control amount is calculated by using a PI controller with the temperature limit value as the reference instruction and the temperature estimation value as the controlled parameter, the engine fuel flow is corrected online, and the gas turbine outlet temperature adaptive limit protection control is realized.

2. The turbine outlet temperature adaptive limit protection method of claim 1, wherein, The step S1 specifically comprises the steps of: S11, obtaining the state parameters of the turboshaft engine, including flight altitude, flight speed, gas turbine speed, power turbine speed, total distance, fuel flow, and air discharge; S12, normalizing the obtained state parameters of the turboshaft engine; S13, fitting the normalized state parameters of the turboshaft engine to obtain the mapping matrix between the input and output, and obtaining the eigenvalue matrix by singular value decomposition.

3. The turbine outlet temperature adaptive limit protection method of claim 2, wherein, Step S12 specifically comprises the steps of: According to the mathematical expectation and standard deviation of the signal, the input and output sample sequences are normalized to unify the evaluation standard: ； wherein X denotes the input data ，X out denotes the normalized input data, μ and σ is X the mean and standard deviation of ε is a small normal number added for division stability.

4. The turbine outlet temperature adaptive limit protection method of claim 3, wherein, S13 specifically comprises the steps of: S131, obtain the input data by least square fitting X to the output data Y mapping matrix: ； S132, obtaining the eigenvalue by decomposing the above formula using singular values: ； In the formula, U M express M orthogonal matrix of order 1, V N express N orthogonal matrix of order, r =rank( C M ) represents the mapping matrix C m rank, κ 1. κ 2…、 κ r for r Each feature value.

5. The turbine outlet temperature adaptive limit protection method of claim 4, wherein, In step S2, the specific strategy for the input-feedback double adaptive long short-term memory network to adaptively allocate the weight of the input gate according to the size of the eigenvalue is: ； In the formula, tanh(•) is the hyperbolic tangent function, W ij An initial weight value corresponding to the i-th input parameter in the input gate of the long short-term memory network, j A weight after adaptive distribution of the i-th input parameter in the input gate, j ​​ The specific strategy for the input-feedback double adaptive long short-term memory network to adaptively correct the feedback weight corresponding to each input parameter in the forgetting gate according to the size of the eigenvalue is: ； In the formula, W fj an initial negative feedback weight value corresponding to an input of a forget gate of a long short-term memory network, j a corresponding negative feedback weight correction value, a corresponding negative feedback weight correction value.

6. The turbine outlet temperature adaptive limit protection method of claim 5, wherein, In step S2, the forward propagation process of the input-feedback double adaptive long short-term memory network is: ； In the formula, f t , i t , o t respectively represent the activation values of the forget gate, the input gate and the output gate, , C t represent the candidate cell state and the current cell state; h t represent the hidden state output; σ represent the sigmoid activation function; and ⊙ represents the Hadamard product, W i , W f , W C are respectively the input, feedback and cell weight matrices.

7. The turbine outlet temperature adaptive limit protection method of claim 6, wherein, In step S2, the specific strategy for the turboshaft engine on-board model is: ； wherein the on-board model of the turboshaft engine is based on the current k time instant and the fuel flow of the previous two time instants k -1, k -2 W f , the collective pitch of the rotor θ 0 , the bleed air flow of the intermediate stage of the compressor L ct , the gas turbine speed of the previous two time instants N g , the power turbine speed N p , the gas turbine outlet temperature T 45 , the flight altitude H and the forward speed V c are inputs; the gas turbine speed of the current k time instant N g , the power turbine speed N p , the engine output torque T qe are outputs, f UAFLSTM (•) is an input-output mapping function of the on-board model of the turboshaft engine based on a long short-term memory network with input-feedback double adaptation.

8. A turbine outlet temperature adaptive limiting protection device characterized by, Comprise: An eigenvalue calculation module for obtaining the state parameters of the turboshaft engine and calculating the corresponding eigenvalue matrix, wherein the eigenvalue is used to represent the correlation between the state parameters and the output, and the size of the eigenvalue is proportional to the correlation; An engine performance prediction module for the on-line prediction of the engine performance parameters by the turboshaft engine on-board model according to the state parameters of the turboshaft engine and inputting to the unscented Kalman state estimator, wherein the turboshaft engine on-board model comprises an input-feedback double adaptive long short-term memory network, wherein the input-feedback double adaptive long short-term memory network adaptively allocates the weight of the input gate according to the size of the eigenvalue, and adaptively corrects the feedback weight corresponding to each input parameter in the forgetting gate according to the size of the eigenvalue; a gas turbine outlet temperature estimation module for the unscented Kalman state estimator to estimate the gas turbine outlet temperature T online from the predicted engine performance parameters 45 ; Fuel flow control module for limiting the gas turbine outlet temperature T 45 When the temperature limit value is exceeded, the control amount is calculated by using a PI controller with the temperature limit value as the reference instruction and the temperature estimation value as the controlled parameter, the fuel flow of the engine is corrected on-line, and the self-adaptive limiting protection control of the gas turbine outlet temperature is realized.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the turboshaft engine on-board model according to the size of the eigenvalue.

10. A storage medium, the storage medium comprising a stored program, characterized in that The program runs to control the device where the storage medium is located to implement the steps of the turboshaft engine on-board model according to the size of the eigenvalue.

Images