A method for modeling a gas turbine start-up process based on test data

Abstract

The application is suitable for the technical field of gas turbine modeling simulation, and provides a gas turbine starting process modeling method based on test data, which comprises the following steps: obtaining measured parameters of a target gas turbine starting process, and performing segmented re-calibration processing on the measured parameters according to preset parameters of a preset gas turbine to obtain calibrated parameters; establishing a Hammerstein model according to the calibrated parameters; obtaining an initial time constant of a sensor model; based on aerodynamic thermodynamic principles, a component-level mechanism model of a target gas turbine slow-speed operation is constructed by using an iterative method, and the Hammerstein model and the component-level mechanism model are fused through a preset filter to obtain a target model; and the initial time constant of the sensor model and the preset time constant of the preset filter in the target model are optimized based on actual test data of the target gas turbine to obtain an optimized model.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of gas turbine modeling simulation, and particularly relates to a modeling method for a gas turbine starting process based on test data. BACKGROUND

[0002] A gas turbine is a complex thermodynamic system with strong nonlinearity, time variation and multivariable coupling, and is composed of a compressor, a combustor, a gas turbine and a power turbine. Finding the corresponding relationship between input and output in the starting process of the gas turbine, establishing a perfect model and performing numerical simulation on the model are important contents of design work. Starting numerical simulation can reduce the risk of physical tests, shorten the development cycle and effectively promote the design and development of new high-efficiency gas turbines in China.

[0003] The research on high-precision simulation technology for the starting process of a gas turbine is still in an insufficient stage at home and abroad. At present, researchers tend to use support vector machines to build a mathematical model of the starting process of a gas turbine, but this method highly depends on a large amount of input and output data for training and optimization. In the face of the starting design of a new gas turbine, due to the lack of targeted data resources, the general practice is to artificially and statically finely adjust the existing starting data based on traditional engineering experience to establish an applicable model. However, this way will cause a certain degree of deviation between the numerical simulation results and the actual test data. SUMMARY

[0004] The embodiment of the application provides a modeling method for a starting process of a gas turbine based on test data, and aims at the problem of deviation between numerical simulation results and actual test data in the prior art.

[0005] The embodiment of the application provides a modeling optimization method for a starting process of a gas turbine based on test data, and comprises the following steps:

[0006] Obtaining a measured parameter of the target gas turbine starting process, and performing segmented re-calibration processing on the measured parameter according to a preset parameter of a preset gas turbine to obtain a calibrated parameter;

[0007] Establishing a Hammerstein model according to the calibrated parameter, wherein the Hammerstein model is used for characterizing the starting process of the target gas turbine, and the Hammerstein model comprises a sensor model used for monitoring an operating parameter of the target gas turbine in the starting process;

[0008] Obtaining an initial time constant of the sensor model;

[0009] Based on the principle of aerodynamic thermodynamics, an iterative method is used to construct a component-level mechanism model of the target gas turbine slow-speed operation, and a preset filter is used to fuse the Hammerstein model and the component-level mechanism model to obtain a target model, wherein a preset time constant of the preset filter is consistent with an initial time constant of the sensor model;

[0010] Based on actual test data of the target gas turbine, the initial time constant of the sensor model in the target model and the preset time constant of the preset filter are optimized to obtain an optimized model.

[0011] Optionally, the step of obtaining the measurement parameters of the target gas turbine startup process and performing segmented re-calibration on the measurement parameters according to preset parameters of a preset gas turbine to obtain calibrated parameters, comprises:

[0012] A key node in the target gas turbine startup process is obtained, and the measurement parameters are segmented according to the key node to obtain pre-ignition measurement parameters and post-ignition measurement parameters, wherein the key node comprises an ignition time, an ignition speed, a slow-speed operation fuel, and a slow-speed operation speed;

[0013] The pre-ignition measurement parameters are time-based data re-calibrated according to a preset pre-ignition measurement parameter sequence in the preset parameters to obtain first measurement parameters, and the formula is as follows:

[0014]

[0015] Wherein is a pre-ignition measurement parameter sequence of the target gas turbine, and the measurement parameter sequence comprises a gas turbine speed sequence, a power turbine speed sequence, a compressor outlet pressure sequence, and a power turbine outlet temperature sequence; is a sensor proportionality coefficient of the pre-ignition measurement parameters of the target gas turbine; is a preset pre-ignition measurement parameter sequence of the preset gas turbine;

[0016] The post-ignition measurement parameters are fuel-based data re-calibrated according to a preset post-ignition measurement parameter sequence in the preset parameters to obtain second measurement parameters, and the formula is as follows:

[0017]

[0018] Wherein is a post-ignition measurement parameter sequence of the target gas turbine; is a sensor proportionality coefficient of the post-ignition measurement parameters; is a post-ignition measurement parameter value of the preset gas turbine; is a post-ignition measurement parameter value of the preset gas turbine; a preset sequence of measured parameters after preset ignition of the preset gas turbine;

[0019] a combination of the first measured parameters and the second measured parameters as the calibration parameters.

[0020] Optionally, the step of establishing a Hammerstein model according to the calibration parameters comprises:

[0021] dividing the starting process of the target gas turbine to obtain a first stage for representing zero speed to ignition speed, a second stage for representing ignition speed to starting motor disengagement, and a third stage for representing starting motor disengagement to slow speed;

[0022] wherein the mathematical representation of the first stage is:

[0023]

[0024] the mathematical representation of the second stage is:

[0025]

[0026] the mathematical representation of the third stage is:

[0027]

[0028] wherein, is the torque provided to the starting motor, is the torque consumed by the gas turbine, is the torque consumed by the compressor, is the mechanical efficiency consumed by friction and accessories, is the angular speed of the gas turbine, is the rotational inertia of the gas generator rotor;

[0029] obtaining a nonlinear function between the control input variables and the calibration parameters, the mathematical representation of the nonlinear function is:

[0030]

[0031] when the gas turbine reaches the slow speed, establishing a linear dynamic system of the target gas turbine, the mathematical representation of the linear dynamic system is:

[0032]

[0033] combining the static nonlinear process and the dynamic linear process, establishing a Hammerstein model according to the calibration parameters and the linear dynamic system, the mathematical representation of the Hammerstein model is:

[0034]

[0035] wherein, is a complex frequency variable after Laplace transform, is a linear dynamic system.

[0036] Optionally, the step of establishing the Hammerstein model according to the calibration parameter further comprises:

[0037] Modeling the sensor in the target gas turbine as a sensor linear dynamic system, and determining the initial time constant of the sensor varying with time by using a nonlinear system identification method and engineering experience;

[0038] Obtaining a first cost function of the difference between the target gas turbine and the preset gas turbine, and the mathematical representation of the first cost function is:

[0039]

[0040] wherein, is the calibration parameter at the i th sampling time, is the control input at the i th sampling time, is the transfer function in the discrete time domain; Using an optimization algorithm to minimize the first cost function to obtain the initial time constant of the target gas turbine varying with time during the starting process;

[0041] Determining the target gas turbine speed corresponding to the initial time constant in the preset corresponding data;

[0042] Using a dynamic scheduling method to obtain an optimal time constant matched with the target gas turbine speed, updating the initial time constant according to the optimal time constant, and obtaining a sensor model for monitoring the operating parameters of the target gas turbine during the starting process.

[0043] Optionally, the step of constructing the component-level mechanism model of the target gas turbine slow-speed process based on the aerodynamic thermodynamic principle and using an iterative method comprises:

[0044] Based on the aerodynamic thermodynamic principle, a component-level mechanism model of the gas turbine slow-speed process is constructed by using an iterative method, and a balance equation of the component-level model is established, wherein the balance equation includes a flow balance equation, a power balance equation and a pressure balance equation:

[0045] The flow balance equation is:

[0046]

[0047] ​​

[0048] wherein, and are the actual inlet flow rates of the gas turbine and the power turbine, respectively, and are the inlet flow rates of the gas turbine and the power turbine, respectively;

[0049] the power balance equation is:

[0050]

[0051] wherein, , , and are the powers of the gas turbine, the power turbine, the compressor and the load shaft, respectively, and are the efficiencies of the gas turbine and the power turbine, respectively;

[0052] the pressure balance equation is:

[0053]

[0054] wherein, and are the ambient pressure and the exit pressure of the tail nozzle, respectively, is the deviation.

[0055] Optionally, before the step of fusing the Hammerstein model and the component-level mechanism model through a preset filter, the method further comprises:

[0056] determining the steady-state values of the mechanism model slow-speed point based on the flow balance equation, the power balance equation and the pressure balance equation;

[0057] taking the slow-speed point corresponding to the steady-state values as a target point, and building a filter corresponding to the sensor model at the target point, wherein the time constant of the filter is consistent with the time constant of the sensor.

[0058] Optionally, the step of optimizing the initial time constant of the sensor model in the target model and the preset time constant of the preset filter based on the actual test data of the target gas turbine comprises:

[0059] determining the fuel sequence in the actual test data and the change rule of the measured data in the actual test data based on a preset difference equation, wherein the mathematical representation of the preset difference equation is:

[0060]

[0061] wherein, is a coefficient related to the time constant, by the transfer function is obtained by transformation, is the result of processing by the filter on the input signal ;

[0062] According to the change rule, the time constant is identified by using a nonlinear least square method, and a second cost function is constructed, and a mathematical representation of the second cost function is:

[0063]

[0064] wherein, is a weight factor;

[0065] The second cost function is minimized by using a Newton method, and the optimized time constant of the sensor model and the optimized time constant of the filter model in the Hammerstein model are obtained by using the coefficient and a preset conversion function, and a mathematical representation of the preset conversion function is:

[0066]

[0067] wherein, is a sampling period, is an optimized time constant.

[0068] Optionally, the method further comprises:

[0069] Segmented identification is performed on actual test data to obtain a time constant sequence and a gas turbine speed sequence corresponding to the time constant sequence;

[0070] Under the same atmospheric temperature and atmospheric pressure as the actual test data, a numerical simulation is performed on the target gas turbine start-up process.

[0071] Optionally, the step of performing the numerical simulation on the target gas turbine start-up process comprises:

[0072] Obtaining change data of the gas turbine speed during the start-up process of the target gas turbine;

[0073] According to the change data, an optimal time constant corresponding to the changed gas turbine speed is dynamically selected.

[0074] Compared with the prior art, the beneficial effects of the embodiments of the present application are that: by acquiring a measurement parameter of a target gas turbine starting process, and performing segmented re-calibration processing on the measurement parameter according to a preset parameter of a preset gas turbine, a calibration parameter is obtained; a Hammerstein model is established according to the calibration parameter, the Hammerstein model is used to characterize the starting process of the target gas turbine, and the Hammerstein model includes a sensor model used to monitor the operating parameter of the target gas turbine in the starting process; an initial time constant of the sensor model is acquired; based on aerodynamic thermodynamic principles, an iterative method is used to construct a component-level mechanism model of a slow vehicle of the target gas turbine, and the Hammerstein model and the component-level mechanism model are fused through a preset filter to obtain a target model, a preset time constant of the preset filter is consistent with the initial time constant of the sensor model; the initial time constant of the sensor model in the target model and the preset time constant of the preset filter are optimized based on actual test data of the target gas turbine to obtain an optimized model. The high-precision dynamic numerical simulation of the starting process of the gas turbine is realized, and the numerical simulation precision of the starting process of the gas turbine is improved. Taking the starting simulation of a certain type of gas turbine as an example, before the application of the present application, the maximum value of the relative error between the key measurement parameters in the starting stage and the actual test data is as high as 8.73%, which may cause some limit conditions to be touched during the actual test to cause emergency shutdown. However, after the application of the present application, the maximum value of the relative error of the key measurement parameters is successfully reduced to 3.88%, especially in that the relative error of the power turbine outlet temperature is greatly reduced by 4.92%, so that the simulation result meets the engineering precision requirement. This improvement not only ensures that the change trend of each measurement parameter in the simulation process is consistent with the actual situation, but also improves the precision of the numerical simulation of the starting process of the gas turbine. BRIEF DESCRIPTION OF DRAWINGS

[0075] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0076] Figure 1 is a gas turbine starting process modeling method flowchart provided by the present application based on test data;

[0077] Figure 2 is another gas turbine starting process modeling method flowchart provided by the present application based on test data;

[0078] Figure 3It is provided by the present application that the gas turbine speed is taken as an example, and the data re-calibration diagram of the gas turbine speed before ignition based on time is shown;

[0079] Figure 4 It is provided by the present application that the gas turbine speed is taken as an example, and the data re-calibration diagram of the gas turbine speed after ignition based on fuel is shown;

[0080] Figure 5 It is provided by the present application that the relationship between the torque and the gas turbine speed during the starting process is shown;

[0081] Figure 6 It is provided by the present application that the Hammerstein model diagram of the gas turbine starting process is shown;

[0082] Figure 7 It is provided by the present application that the data model and mechanism model fusion diagram is shown. DETAILED DESCRIPTION

[0083] In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, technologies, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

[0084] It should be understood that the term "includes" when used in the specification and the appended claims herein, specifies the presence of stated features, integers, steps, operations, elements, and / or components but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0085] It should also be understood that the term "and / or" when used in the specification and the appended claims herein, means and encompasses any and all possible combinations of one or more of the associated listed items and can be used interchangeably with the term "or".

[0086] As used in this specification and any claims of this application, the terms "if" and "when" can be construed to mean "when" or "if," depending on the context. Similarly, the phrase "if it is determined" or "if [a described condition or event] is detected" can be construed to mean "if a determination is made" or "in response to a determination," or "if [the described condition or event] is detected," depending on the context.

[0087] In addition, in the description of the present application and the appended claims, the terms "first", "second", "third", etc. are used only to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0088] In the present application, the reference to "one embodiment" or "some embodiments" means that the specific features, structures or characteristics described in connection with the embodiment are included in one or more embodiments of the present application. Therefore, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in further some embodiments" and the like appearing in the present specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "include", "contain", "have" and their variants mean "including but not limited to", unless otherwise specifically emphasized.

[0089] As Figure 1 shown, the embodiment of the present application provides a gas turbine starting process modeling optimization method based on test data, comprising the following steps:

[0090] S101, obtaining the measurement parameters of the target gas turbine starting process, and performing segmented re-calibration processing on the measurement parameters according to the preset parameters of the preset gas turbine to obtain calibrated parameters;

[0091] S102, establishing a Hammerstein model according to the calibrated parameters, the Hammerstein model being used to characterize the starting process of the target gas turbine, the Hammerstein model including a sensor model for monitoring the operating parameters of the target gas turbine during the starting process;

[0092] S103, obtaining the initial time constant of the sensor model;

[0093] S104, based on the principles of aerodynamic thermodynamics, using an iterative method to construct a component-level mechanism model of the slow vehicle of the target gas turbine, and fusing the Hammerstein model and the component-level mechanism model through a preset filter to obtain a target model, the preset time constant of the preset filter being consistent with the initial time constant of the sensor model;

[0094] S105, optimizing the initial time constant of the sensor model in the target model and the preset time constant of the preset filter based on the actual test data of the target gas turbine to obtain an optimized model.

[0095] In a possible implementation, as Figure 2As shown, the embodiment of the present application provides another modeling optimization method for gas turbine starting process based on test data, which comprises the following steps: S1, obtaining the measured parameters of the existing gas turbine starting process, and analyzing the ignition point and slow speed point of the gas turbine, wherein the measured parameters are turbine speed , power turbine speed , compressor outlet pressure , power turbine outlet temperature , fuel flow , ignition time and starting time ; combining the design requirements of a new type of gas turbine (preset parameters of the preset gas turbine) to perform different degrees of segmented data re-calibration on the measured parameters of the existing gas turbine starting process (equivalent to obtaining the measured parameters of the target gas turbine starting process, and performing segmented re-calibration processing on the measured parameters according to the preset parameters of the preset gas turbine to obtain calibrated parameters);

[0096] S2, establishing the gas turbine starting data model, and the relationship between torque and turbine speed during the starting process is as shown in Figure 5 . In the first stage of the starting process, the gas turbine is completely driven by the starting motor to accelerate the gas generator rotor until the combustion chamber ignites, at which time the remaining power is the difference between the torque generated by the starting motor and the cold running consumption torque, which is called the ignition speed; in the second stage, the starting motor and the gas turbine jointly work to accelerate the engine shaft, at which time the remaining power is the difference between the sum of the torques of the starting motor and the gas turbine and the compressor consumption power, which is called the decoupling speed; in the third stage, the starting motor is decoupled and the gas turbine alone drives the engine rotor to accelerate until the slow speed speed is reached, at which time the remaining power is the difference between the gas turbine power and the compressor consumption power. Taking fuel as the control input variable, combining the nonlinear relationship between the starting process of the gas turbine and the measured parameters after segmented data re-calibration and the dynamic characteristics when reaching the slow speed speed, a Hammerstein data model as shown in Figure 6 is established (equivalent to establishing a Hammerstein model according to the calibrated parameters, and the Hammerstein model is used to characterize the starting process of the target gas turbine);

[0097] S3, building a model of the temperature sensor, pressure sensor and speed sensor of the gas turbine, each sensor corresponds to a different transfer function (equivalent to the Hammerstein model comprising a sensor model for monitoring the operating parameters of the target gas turbine during the starting process), as shown in Figure 6The dynamic linear link shown in the middle. Analysis of gas turbine speed changes on the compressor pressure ratio, the mixture concentration and the combustion rate in the combustion chamber; using parameter identification method to get the time-varying time constant of the pressure sensor and the speed sensor in the starting process of the gas turbine, and the time-varying time constant of the temperature sensor in the starting process of the gas turbine is obtained based on engineering experience; each sensor has a sequence of time constants and a corresponding sequence of gas turbine speeds, and the optimal time constant matching the current speed is obtained by using dynamic scheduling method; due to the lack of targeted data resources, the temperature sensor is generally selected as 8~10 according to experience, that is ; The time constant of the pressure sensor is consistent with the speed sensor, that is (corresponding to obtaining the initial time constant of the sensor model);

[0098] Step S4, based on the aerodynamic thermodynamic principle of gas turbine operation, an iterative method is used to construct a mechanism model of the components above the slow speed of the gas turbine. The slow speed fuel value is known, and when the deviations of the flow balance equation, the power balance equation and the pressure balance equation are within the specified range, the steady-state value of the mechanism model at the slow speed point is calculated. A filter is built at the switching point to integrate the data model and the mechanism model to avoid numerical jumps when the model switches and touches the gas turbine starting limit condition, at this time the time constant of the filter is consistent with the time constant of the sensor in the starting process, and the integrated model as shown in Figure 7 is obtained (corresponding to based on the aerodynamic thermodynamic principle, using an iterative method to construct a mechanism model of the components above the slow speed of the target gas turbine, and integrating the Hammerstein model and the mechanism model through a preset filter to obtain a target model, the preset time constant of the preset filter is consistent with the initial time constant of the sensor model);

[0099] Step S5, the variation of the measured parameters is the result of the input fuel sequence processed by the sensor, and the time constants of each sensor are generated and optimized by using nonlinear least squares method and Newton method (corresponding to based on the actual test data of the target gas turbine, the initial time constant of the sensor model in the target model and the preset time constant of the preset filter are optimized to obtain an optimized model), at this time the time constants corresponding to the four sensors are inconsistent, that is ; The actual test data is segmented and identified to obtain a sequence of time constants and a corresponding sequence of gas turbine speeds, and the optimal time constant corresponding to the current speed is dynamically selected when the gas turbine speed changes in the starting process to improve the simulation accuracy.

[0100] By acquiring measurement parameters of the target gas turbine startup process and performing segmented recalibration on these parameters according to preset parameters of the gas turbine, calibration parameters are obtained. A Hammerstein model is established based on these calibration parameters. This Hammerstein model characterizes the startup process of the target gas turbine and includes sensor models for monitoring the operating parameters of the target gas turbine during startup. The initial time constant of the sensor models is obtained. Based on the principles of aerothermodynamics, an iterative method is used to construct a component-level mechanism model of the target gas turbine at idle speed. The Hammerstein model and the component-level mechanism model are then fused using a preset filter to obtain the target model. The preset time constant of the preset filter is consistent with the initial time constant of the sensor models. Based on the actual test data of the target gas turbine, the initial time constant of the sensor models and the preset time constant of the preset filter in the target model are optimized to obtain an optimized model. This achieves high-precision dynamic numerical simulation of the gas turbine startup process, improving the accuracy of the numerical simulation of the gas turbine startup process. Taking the startup simulation of a specific gas turbine model as an example, before adopting this invention, the maximum relative error between the key measurement parameters during the startup phase and the actual test data was as high as 8.73%, which could lead to emergency shutdowns due to certain limiting conditions encountered during actual testing. However, after applying this invention, the maximum relative error of the key measurement parameters was successfully reduced to 3.88%, particularly with a significant reduction of 4.92% in the relative error of the power turbine outlet temperature, ensuring that the simulation results meet engineering accuracy requirements. This improvement not only ensures that the changing trends of each measurement parameter during the simulation are consistent with reality but also improves the accuracy of the numerical simulation of the gas turbine startup process.

[0101] In one possible implementation, the step of acquiring measurement parameters of the target gas turbine startup process and performing segmented recalibration processing on the measurement parameters according to preset parameters of the gas turbine to obtain calibration parameters includes:

[0102] The key nodes in the startup process of the target gas turbine are obtained, and the measurement parameters are segmented according to the key nodes to obtain the measurement parameters before ignition and the measurement parameters after ignition. The key nodes include ignition time, ignition speed, idle fuel and idle speed.

[0103] like Figure 3 As shown, the pre-ignition measurement parameters are recalibrated based on time according to the preset pre-ignition measurement parameter sequence in the preset parameters to obtain the first measurement parameter, as shown in the following formula:

[0104]

[0105] wherein is a sequence of pre-ignition measured parameters of the target gas turbine, the sequence of measured parameters comprising a sequence of gas turbine speed, a sequence of power turbine speed, a sequence of compressor outlet pressure and a sequence of power turbine outlet temperature; is a pre-ignition measured parameter sensor scaling factor of the target gas turbine; is a sequence of pre-ignition measured parameters of the preset gas turbine;

[0106] As Figure 4 shown, the post-ignition measured parameters are fuel-based data re-calibrated according to a sequence of preset post-ignition measured parameters in the preset parameters, to obtain a second measured parameter, the formula being as follows:

[0107]

[0108] wherein is a sequence of post-ignition measured parameters of the target gas turbine; is a post-ignition measured parameter sensor scaling factor; is a measured parameter value at the ignition of the preset gas turbine; is a measured parameter value at the ignition of the preset gas turbine; is a sequence of preset post-ignition measured parameters of the preset gas turbine;

[0109] The collection of the first measured parameter and the second measured parameter is taken as the calibration parameter.

[0110] In a possible implementation, the step of establishing a Hammerstein model according to the calibration parameter comprises:

[0111] The starting process of the target gas turbine is divided to obtain a first stage for characterizing zero speed to ignition speed, a second stage for characterizing ignition speed to starting motor disengagement, and a third stage for characterizing starting motor disengagement to slow speed;

[0112] wherein the mathematical representation of the first stage is:

[0113]

[0114] The mathematical representation of the second stage is:

[0115]

[0116] The mathematical representation of the third stage is:

[0117]

[0118] wherein, The torque provided to start the motor The torque consumed by the gas turbine, The torque consumed by the compressor. Mechanical efficiency due to friction and accessory consumption. Let be the angular velocity of the gas turbine. The moment of inertia of the gas generator rotor;

[0119] Obtain the nonlinear function between the control input variable and the calibration parameter, wherein the mathematical representation of the nonlinear function is:

[0120]

[0121] Once the gas turbine reaches idle speed, a linear dynamic system of the target gas turbine is established, and the mathematical representation of the linear dynamic system is as follows:

[0122]

[0123] Combining static nonlinear processes and dynamic linear processes, a Hammerstein model is established based on the calibration parameters and the linear dynamic system. The mathematical representation of the Hammerstein model is as follows:

[0124]

[0125] in, The complex frequency variable is the result of the Laplace transform. It is a linear dynamic system.

[0126] In one possible implementation, the step of establishing the Hammerstein model based on the calibration parameters further includes:

[0127] The sensors inside the target gas turbine are modeled as a linear dynamic system, and the time-varying initial time constant of the sensors is determined using nonlinear system identification methods and engineering experience.

[0128] A first cost function is obtained to determine the difference between the target gas turbine and the preset gas turbine. The mathematical representation of the first cost function is as follows:

[0129]

[0130] in, For the first The calibration parameters at each sampling time. For the first Control input at each sampling time, The transfer function in the discrete time domain;

[0131] minimizing the first cost function using an optimization algorithm to obtain an initial time constant of the target gas turbine during the start-up process;

[0132] determining a target gas turbine speed corresponding to the initial time constant in preset corresponding data;

[0133] obtaining an optimal time constant matched with the target gas turbine speed by using a dynamic scheduling method, updating the initial time constant according to the optimal time constant to obtain a sensor model for monitoring an operating parameter of the target gas turbine during the start-up process.

[0134] In a possible implementation, the step of constructing the component-level mechanism model of the target gas turbine slow-speed operation based on aerodynamic thermodynamic principles and using an iterative method includes:

[0135] constructing a component-level mechanism model of the gas turbine slow-speed operation based on aerodynamic thermodynamic principles and using an iterative method, and establishing a balance equation of the component-level model, the balance equation including a flow balance equation, a power balance equation, and a pressure balance equation:

[0136] The flow balance equation is:

[0137]

[0138] wherein, and are actual inlet flow rates of the gas turbine and the power turbine, respectively, and are inlet flow rates of the gas turbine and the power turbine, respectively;

[0139] The power balance equation is:

[0140]

[0141] wherein, , , and are powers of the gas turbine, the power turbine, the compressor, and the load shaft, respectively, and are efficiencies of the gas turbine and the power turbine, respectively;

[0142] The pressure balance equation is:

[0143]

[0144] wherein, and are ambient pressure and tail jet outlet pressure, respectively, is a deviation.

[0145] In a possible implementation, before the step of fusing the Hammerstein model and the component-level mechanism model through the preset filter, the method further comprises:

[0146] determining a steady-state value of a mechanism model slow point based on the flow balance equation, the power balance equation and the pressure balance equation;

[0147] taking the slow point corresponding to the steady-state value as a target point, and building a filter corresponding to the sensor model at the target point, a time constant of the filter being consistent with a time constant of the sensor.

[0148] In a possible implementation, the step of optimizing the initial time constant of the sensor model in the target model and the preset time constant of the preset filter based on the actual test data of the target gas turbine comprises:

[0149] determining a variation law of the fuel sequence in the actual test data and the measured data in the actual test data based on a preset difference equation, the preset difference equation being mathematically represented as:

[0150]

[0151] wherein, is a coefficient related to the time constant, is obtained by a transfer function is obtained by transformation, is a result processed by the filter on the input signal ;

[0152] according to the variation law, identifying the time constant by using a nonlinear least square method, and constructing a second cost function, the second cost function being mathematically represented as:

[0153]

[0154] wherein, is a weight factor;

[0155] minimizing the second cost function by using a Newton method, and obtaining the optimized time constant of the sensor model in the Hammerstein model and the optimized time constant of the filter model by using the coefficient and a preset conversion function, the preset conversion function being mathematically represented as:

[0156]

[0157] wherein, is a sampling period, is the optimized time constant.

[0158] In a possible implementation, the method further comprises:

[0159] segmentally identifying the actual test data to obtain a time constant sequence and a gas turbine speed sequence corresponding to the time constant sequence;

[0160] numerically simulating the target gas turbine starting process under the same atmospheric temperature and atmospheric pressure as the actual test data.

[0161] In a possible implementation, the step of numerically simulating the target gas turbine starting process comprises:

[0162] obtaining the change data of the gas turbine speed of the target gas turbine during the starting process;

[0163] dynamically selecting the optimal time constant corresponding to the changed gas turbine speed according to the change data.

[0164] For example, for a certain type of gas turbine (target gas turbine), under the initial condition of keeping the same atmospheric temperature and atmospheric pressure as the actual test data, the model is simulated until the gas turbine speed reaches the slow speed point. When the actual test data is not used for optimization, the maximum value of the relative error between the key measurement parameters and the actual test data is as high as 8.73%, and the power turbine outlet temperature changes greatly at the model fusion, which may cause the actual test to touch some limit conditions and cause emergency shutdown. However, after optimization using the actual test data, the maximum value of the relative error of the key measurement parameters is successfully reduced to 3.88%, especially the relative error of the power turbine outlet temperature is greatly reduced by 4.92%, and the simulation result meets the engineering precision requirement; before ignition, the power turbine speed remains zero, and the power turbine outlet temperature keeps consistent with the atmospheric temperature during the test; at the time of ignition, the gas turbine speed increases, the power turbine speed starts to increase from zero, and the power turbine outlet temperature significantly increases; when approaching the slow fuel, the change rates of the gas turbine speed and the power turbine outlet temperature are both slow, which conforms to the mechanism characteristics of the actual starting process of the gas turbine.

[0165] Those skilled in the art can appreciate that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0166] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / network device and method can be implemented in other manners. For example, the embodiments of the apparatus / network device described above are merely illustrative. For example, the division of the modules or units is merely logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0167] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0168] The above-described embodiments are merely used to illustrate the technical solutions of the present application, but not limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalent replacements; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for modeling and optimization of a gas turbine start-up process based on test data, characterized in that, The method comprises the following steps: obtaining measurement parameters of the target gas turbine starting process, and performing segmented re-calibration processing on the measurement parameters according to preset parameters of a preset gas turbine to obtain calibrated parameters; establishing a Hammerstein model according to the calibrated parameters, the Hammerstein model being used to characterize the starting process of the target gas turbine, and the Hammerstein model comprising a sensor model used to monitor operating parameters of the target gas turbine during the starting process; the step of establishing the Hammerstein model according to the calibrated parameters comprises: dividing the starting process of the target gas turbine to obtain a first stage used to characterize zero speed to ignition speed, a second stage used to characterize ignition speed to starting motor disengagement, and a third stage used to characterize starting motor disengagement to slow-speed speed; the mathematical representation of the first stage is as follows: the mathematical representation of the second stage is as follows: the mathematical representation of the third stage is as follows: wherein, torque provided to start the electric machine, torque consumed by the gas turbine, torque consumed by the compressor, mechanical efficiency consumed by friction and accessories, angular velocity of the gas turbine, moment of inertia of the gas generator rotor; obtaining a nonlinear function between a control input variable and the calibrated parameters, the mathematical representation of the nonlinear function being as follows: after the gas turbine reaches the slow-speed speed, establishing a linear dynamic system of the target gas turbine, the mathematical representation of the linear dynamic system being as follows: integrating the static nonlinear process and the dynamic linear process, establishing a Hammerstein model according to the calibrated parameters and the linear dynamic system, and the mathematical representation of the Hammerstein model being as follows: wherein is the complex frequency variable after Laplace transformation, is a linear dynamic system; obtaining an initial time constant of the sensor model; based on aerodynamic thermodynamic principles, constructing a component-level mechanism model of the slow-speed speed of the target gas turbine by using an iterative method, and fusing the Hammerstein model and the component-level mechanism model through a preset filter to obtain a target model, the preset time constant of the preset filter being consistent with the initial time constant of the sensor model; optimizing the initial time constant of the sensor model in the target model and the preset time constant of the preset filter based on actual test data of the target gas turbine to obtain an optimized model.

2. A method of modeling and optimizing a gas turbine start-up process based on test data according to claim 1, characterized in that, the step of obtaining measurement parameters of the target gas turbine starting process, and performing segmented re-calibration processing on the measurement parameters according to preset parameters of a preset gas turbine to obtain calibrated parameters comprises: obtaining key nodes in the starting process of the target gas turbine, and segmenting the measurement parameters according to the key nodes to obtain measurement parameters before ignition and measurement parameters after ignition, the key nodes comprising ignition time, ignition speed, slow-speed fuel, and slow-speed speed; performing time-based data re-calibration on the measurement parameters before ignition according to a preset measurement parameter sequence before ignition in the preset parameters to obtain first measurement parameters, the formula being as follows: wherein a sequence of measured parameters before the target gas turbine is ignited, the sequence of measured parameters comprising a sequence of gas turbine rotational speed, a sequence of power turbine rotational speed, a sequence of compressor outlet pressure and a sequence of power turbine outlet temperature; a scaling factor of the sequence of measured parameters before the target gas turbine is ignited; a sequence of measured parameters before the target gas turbine is ignited, the sequence of measured parameters comprising a sequence of gas turbine rotational speed, a sequence of power turbine rotational speed, a sequence of compressor outlet pressure and a sequence of power turbine outlet temperature; performing fuel-based data re-calibration on the measurement parameters after ignition according to a preset measurement parameter sequence after ignition in the preset parameters to obtain second measurement parameters, the formula being as follows: wherein is a measured parameter sequence after ignition for the target gas turbine; is a measured parameter sensor scaling factor after ignition; is a measured parameter value at the ignition for the preset gas turbine; is a measured parameter value at the ignition for the preset gas turbine; is a preset measured parameter sequence after ignition for the preset gas turbine; taking the union of the first measurement parameters and the second measurement parameters as the calibrated parameters.

3. A method of modeling a gas turbine start-up process based on test data as recited in claim 1, wherein, The step of establishing the Hammerstein model according to the calibration parameters further comprises: modeling the sensors in the target gas turbine as sensor linear dynamic systems, and determining initial time constants of the sensors that vary with time by using a nonlinear system identification method and engineering experience; obtaining a first cost function of differences between the target gas turbine and the preset gas turbine, the first cost function being mathematically represented as: wherein, is the calibrated parameter at the th sampling instant, is the control input at the th sampling instant, is the transfer function in discrete time domain; minimizing the first cost function by using an optimization algorithm to obtain initial time constants of the target gas turbine that vary with time during a starting process of the target gas turbine; determining a target gas turbine speed corresponding to the initial time constants in preset corresponding data; obtaining optimal time constants matched with the target gas turbine speed by using a dynamic scheduling method, updating the initial time constants according to the optimal time constants, and obtaining a sensor model for monitoring operating parameters of the target gas turbine during the starting process.

4. A method of modeling and optimizing a gas turbine start-up process based on test data according to claim 1, characterized in that, The step of constructing the component-level mechanism model of the target gas turbine slow-speed operation based on aerodynamic thermodynamic principles and by using an iterative method comprises: constructing a component-level mechanism model of the gas turbine slow-speed operation based on aerodynamic thermodynamic principles and by using an iterative method, and establishing balance equations of the component-level model, the balance equations including a flow balance equation, a power balance equation, and a pressure balance equation: the flow balance equation is: wherein, and are the actual inlet flow rates of the gas turbine and the power turbine, respectively, and are the inlet flow rates of the gas turbine and the power turbine, respectively. the power balance equation is: wherein , , and are the power of the gas turbine, the power turbine, the compressor and the load shaft, respectively, and are the efficiency of the gas turbine and the power turbine, respectively; the pressure balance equation is: wherein, and Penvand Ptailare the ambient pressure and the tailpipe exit pressure, respectively, is the deviation.

5. A method of modeling and optimizing a gas turbine start-up process based on test data according to claim 4, characterized in that, The step of fusing the Hammerstein model and the component-level mechanism model by using a preset filter further comprises: determining steady-state values of the mechanism model slow-speed operation point based on the flow balance equation, the power balance equation, and the pressure balance equation; taking the slow-speed operation point corresponding to the steady-state values as a target point, and building a filter corresponding to the sensor model at the target point, a time constant of the filter being consistent with a time constant of the sensor.

6. A method of modeling and optimizing a gas turbine start-up process based on test data according to claim 1, characterized in that, The step of optimizing the initial time constant of the sensor model and the preset time constant of the preset filter in the target model based on actual test data of the target gas turbine comprises: determining a fuel sequence in the actual test data and a change rule of measured data in the actual test data based on a preset difference equation, the preset difference equation being mathematically represented as: wherein is a coefficient related to a time constant, by a transfer function is transformed to is the result of the processing of the input signal by the filter identifying time constants by using a nonlinear least square method according to the change rule, and constructing a second cost function, the second cost function being mathematically represented as: wherein is a weight factor; minimizing said second cost function by Newton method and using the coefficients and a preset conversion function, a mathematical representation of which is: T = T0 + T1 * (1 - e-a) wherein is a sampling period, is an optimization time constant.

7. A method of modeling a gas turbine start-up process based on test data as recited in claim 1, wherein, The method further comprises: segmenting and identifying the actual test data to obtain a time constant sequence and a gas turbine speed sequence corresponding to the time constant sequence; numerically simulating the starting process of the target gas turbine under the same atmospheric temperature and atmospheric pressure as the actual test data.

8. A method of modeling a gas turbine start-up process based on test data according to claim 7, characterized in that, The step of numerically simulating the starting process of the target gas turbine comprises: obtaining change data of the gas turbine speed during the starting process of the target gas turbine; dynamically selecting optimal time constants corresponding to the changed gas turbine speed according to the change data.

Images