CN117521506A - Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model - Google Patents

Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model Download PDF

Info

Publication number
CN117521506A
CN117521506A CN202311493674.5A CN202311493674A CN117521506A CN 117521506 A CN117521506 A CN 117521506A CN 202311493674 A CN202311493674 A CN 202311493674A CN 117521506 A CN117521506 A CN 117521506A
Authority
CN
China
Prior art keywords
compressor
gas
flow
gas turbine
local model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311493674.5A
Other languages
Chinese (zh)
Inventor
周文祥
谢雨扬
吴广昊
彭文辉
白宇
黄金泉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202311493674.5A priority Critical patent/CN117521506A/en
Publication of CN117521506A publication Critical patent/CN117521506A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M15/00—Testing of engines
    • G01M15/14—Testing gas-turbine engines or jet-propulsion engines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G06F17/12—Simultaneous equations, e.g. systems of linear equations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computational Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Probability & Statistics with Applications (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Geometry (AREA)
  • Chemical & Material Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Medical Informatics (AREA)
  • Operations Research (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Control Of Positive-Displacement Air Blowers (AREA)

Abstract

The invention discloses a turbine rotating speed signal reconstruction method of a turboshaft engine based on a local model. Firstly, through turboshaft engine flow path analysis, selecting parts and sensor information related to gas turbine rotation speed ng calculation, and establishing a ng local model by using a aerodynamic thermodynamic method. When the ng signal does not have faults, the model enters a training mode, decoupling of health parameters is achieved through theoretical analysis, appropriate health parameters are selected, the health parameters of the part are estimated based on a method for solving a nonlinear equation and K-means clustering according to errors of model calculation results and measured data, and the health parameters are introduced into the ng local model as adjustable parameters to complete online correction of the local model. The result shows that the steady state error of the ng reconstruction signal is not more than 0.25%, and the transition state error is not more than 0.33%, which indicates that the signal reconstruction method provided by the invention has high precision. The method does not need to add an additional sensor, and has good real-time performance and strong engineering practicability.

Description

Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model
Technical Field
The invention relates to the technical field of aero-engines, in particular to a turbine shaft engine gas turbine rotating speed signal reconstruction method based on a local model.
Background
The sensor is an important component of the turboshaft engine control system, but the failure rate is high due to the severe working environment. In the control system, the sensor faults account for more than 80% of the total fault number, and the gas turbine rotating speed sensor is one of the most critical input parameters of the turboshaft engine control system, and the faults not only can cause the whole cascade control loop to be incapable of being closed, but also can influence the guide vane position control precision, thereby further causing the engine surge. Therefore, it is necessary to propose a practical gas turbine rotational speed signal reconstruction method to ensure the stability and safety of the operation of the engine control system.
At present, three main research ideas for sensor signal reconstruction are that a high-precision airborne model is utilized, wherein the high-precision airborne model comprises an interpolation model, a component level model or a model-based filter, an observer and other methods are used for reconstructing fault signals; the method fully utilizes other effective sensor information, and provides resolution redundancy through a data driving method such as a neural network, an extreme learning machine, or a relationship fitting method; in recent years, a new thought of combining a model and data driving appears, but at present, the research of the technology of an on-board model in China is still immature, an interpolation model and a component level model cannot fully reflect the individual difference and performance degradation of an engine, methods such as a model-based filter and an observer face the problem of insufficient instantaneity, the data driving method depends on a large amount of fault data, the requirement on hardware of on-board equipment is high, and engineering application is limited. Therefore, a method for reconstructing the gas turbine rotational speed signal of the turboshaft engine with strong engineering practicability is necessary to be developed.
Disclosure of Invention
The invention aims to: aiming at the problems in the background technology, the invention combines a high-precision component level model with effective sensor signals, establishes a local model of the rotating speed of the gas turbine of the turboshaft engine based on a component level modeling thought, designs a training mode to estimate health parameters, improves the prediction precision of the model on the rotating speed signal of the gas turbine, has simple method and is easy to engineering and practical.
The technical scheme is as follows: in order to achieve the above purpose, the invention adopts the following technical scheme:
a turbine rotating speed signal reconstruction method of a turboshaft engine based on a local model comprises the following steps:
step A, establishing the rotation speed n of the gas turbine g A local model;
step B, when the rotating speed n g When no fault occurs, the gas turbine speed n g The local model enters a training mode, decoupling of two health parameters, namely the flow and the efficiency of the gas compressor in the local model, is realized through theoretical analysis, then the health parameter with the greatest influence on the local model is selected, the health parameter of the gas compressor component is estimated based on a method for solving a nonlinear equation and K-means clustering according to the error between the calculation result and the measured data of the local model, and then the health parameter is used as an adjustable parameter to be introduced into the local model, so that the online correction of the local model is completed;
step C, when n g When the signal fails, the local model enters a reconstruction mode, and the corrected local model is used for calculating high-precision n g Reconstructing the signal.
Preferably, the gas turbine rotational speed partial model establishment process in step a is as follows:
step A1, bleed air system Sigma W j (k) Modeling:
bleed air flow Sigma W j (k) Comprising the following steps:
intermediate stage of compressor to gas turbine inlet cooling gas W gt1 (k) Compressor intermediate stage to gas turbine outlet cooling gas W gt2 (k) Cooling gas W from outlet of gas compressor to inlet of power turbine pt1 (k) The cooling air W from the outlet of the air compressor to the outlet of the power turbine pt2 (k) Other intermediate stage bleed air W los (k) The other intermediate stage bleed air comprises air bleed air of an intermediate stage of a compressor, and the bleed air is sealed pneumatically:
∑W j (k)＝W gt1 (k)+W gt2 (k)+W pt1 (k)+W pt2 (k)+W los (k) (1)
step A2, modeling a compressor:
at time k, bleed air specific enthalpy h of outlet of compressor gt1 (k) Specific enthalpy of inlet air flow h of air compressor 2 (k) Specific enthalpy of compressor outlet flow h 3 (k) The following relationship exists:
h gt1 (k)＝h gt2 (k)＝h 3 (k) (2)
intermediate stage bleed air specific enthalpy h of compressor at moment k cpm (k) And gas turbine speed n g (k) Specific enthalpy of inlet air flow h of air compressor 2 (k) Specific enthalpy of compressor outlet flow h 3 (k) The method meets the following conditions:
h pt1 (k)＝h pt2 (k)＝h los (k)＝h cpm (k)＝f m,h (n g (k),h 2 (k),h 3 (k)) (3)
in the formula, h pt1 (k) Represents the specific enthalpy of cooling gas at the inlet of the power turbine, h pt2 (k) Represents the specific enthalpy of cooling gas at the outlet of the power turbine, h los (k) Representing the specific enthalpy of other intermediate stage bleed air; h is a cpm (k) By the gas turbine speed n at time k g (k) Specific enthalpy h of inlet air flow of compressor 2 (k) And compressor outlet air flow specific enthalpy h 3 (k) Calculated, the process uses f m,h A function representation;
according to the gas turbine rotation speed n at time k g (k) Design speed n of gas turbine g,d Total inlet temperature T of air compressor t2 (k) Total temperature T of inlet design point of air compressor t2,d Total pressure P of inlet of compressor t2 (k) And the total pressure P of the outlet of the air compressor t3 (k) Obtaining the converted rotating speed n of the air compressor g,cor (k) Sum pressure ratio pi cp (k)：
Wherein: the subscript d represents the design point parameter of the inlet of the compressor, and the subscript cor represents the conversion parameter;
by compressor flow characteristics f W,cp Obtaining the inlet flow W of the air compressor 2 (k) By compressor efficiency characteristics f η,cp Obtaining the efficiency eta of the compressor cp (k)：
η cp (k)＝f η,cp (n g,cor (k),π cp (k)) (7)
Calculating the specific enthalpy h of the inlet air flow of the air compressor by a fixed specific heat method 2 (k) Specific enthalpy of compressor outlet flow h 3 (k)：
Wherein, h, s and f respectively represent specific enthalpy, specific entropy and oil-gas ratio, g T2h 、g T2s 、g s2h And g h2T As a function, the corner mark 2 represents the compressor inlet, 3 represents the compressor outlet, i represents the flow loss of gas, T t3 (k) The total inlet temperature of the air compressor;
considering the bleed air of the compressor, the outlet flow of the compressor is as follows:
W 3 (k)＝W 2 (k)-∑W j (k) (9)
the power required by the compressor is as follows:
step A3, modeling a combustion chamber:
the combustor components need only establish energy and flow transfer equations, where the energy transfer equations are:
the flow transfer equation the transfer equation is:
W 4 (k)＝W 3 (k)+W f (k) (12)
wherein H is u Is the low heat value eta of the fuel oil b For combustion efficiency, 4 denotes the combustion chamber outlet cross section, h 4 (k) Indicating the specific enthalpy of the outlet airflow of the combustion chamber at the moment k, W f (k) The fuel flow of the combustion chamber at the moment k is represented;
step A4, turbine modeling:
the energy exchange process of the gas turbine inlet cooling gas blending process is represented as:
the flow exchange process of the gas turbine inlet cooling gas blending process is represented as:
W 41 (k)＝W 4 (k)+W gt1 (k) (14)
in the formula, h 41 (k) Represents the section airflow specific enthalpy after mixing the inlet of the gas turbine at the moment k, W 41 (k) Representing the cross-sectional gas flow rate after blending at the gas turbine inlet; and has
W 43 (k)＝W 41 (k) (15)
The energy exchange process of the gas turbine outlet cooling gas blending process is represented as:
the flow exchange process of the gas turbine outlet cooling gas blending process is represented as:
W 44 (k)＝W 43 (k)+W gt2 (k) (17)
in which W is 43 (k) Represents the gas flow rate of the cross section before blending of the gas turbine outlet, h 43 (k) Representing the specific enthalpy of the section airflow before mixing of the outlet of the gas turbine at the moment k, W 44 (k) Representing the gas flow rate after mixing at the outlet of the gas turbine at the time k;
the power turbine inlet cooling gas blending energy exchange process is expressed as:
the power turbine inlet cooling gas blending flow rate exchange process is expressed as:
W 45 (k)＝W 44 (k)+W pt1 (k) (19)
in the formula, h 45 (k) Represents the specific enthalpy of the section airflow before blending of the inlet of the power turbine at the moment k, W 45 (k) The gas flow rate after mixing at the inlet of the power turbine at the moment k is shown;
the gas turbine emits power as follows:
L gt (k)＝W 41 (k)h 41 (k)-W 43 (k)h 43 (k) (20)
step A5, rotor dynamics model:
the residual power of the core machine is expressed as follows according to the principle of the engine
ΔL(k)＝η mgt L gt (k)-L cp (k) (21)
Wherein: η (eta) mgt For rotor mechanical efficiency, L gt (k) Generating power for a gas turbine, L cp (k) The power required by the compressor;
substituting equations (1) - (20) into equation (21) to calculate the residual power, and finally calculating the next-time rotational speed n of the gas turbine g (k+1):
Wherein: ΔL (k) is rotor residual power, T dstep For the simulation step, J is the rotor moment of inertia.
The local model comprises a constructed air entraining system Sigma W i (k) Compressor, combustor, turbine, and rotor dynamics model.
Preferably, in the step B, the decoupling process of the two health parameters of the flow and the efficiency of the compressor in the local model is implemented through theoretical analysis as follows:
step B1.1, since the decoupling process does not require the construction of a complex local model in step A, i.e. the shaft mechanical efficiency η is not taken into account mgt And the gas is discharged by the gas compressor, and then the local model is simplified, so that the power required by the gas compressor is as follows:
L cp (k)＝W 3 (k)h 3 (k)-W 2 (k)h 2 (k) (23)
the gas turbine emits power as follows:
L gt (k)＝W 4 (k)h 4 (k)-W 45 (k)h 45 (k) (24)
the combustor energy exchange is expressed as:
the remaining power of the core machine is as follows:
ΔL(k)＝L gt (k)-L cp (k) (26)
substituting equations (23) - (25) into equation (26), the core residual power is written as:
ΔL(k)＝W f (k)H u (k)η b -W f (k)h 45 (k)+W 2 (k)(h 2 (k)-h 45 (k)) (27)
step B1.2, analytical method(27) It is known that the only healthy parameter affecting the residual power ΔL (k) in the compressor characteristics is the compressor flow W 2 (k) In the modeling of step A, the mechanical efficiency η of the shaft is taken into account mgt And the bleed air of the compressor due to eta mgt And the flow of the gas discharged by the gas compressor is ignored and is close to 1, so that the influence of the efficiency of the gas compressor on the whole calculation accuracy of the local model is small and is ignored, and the decoupling of the health parameters of the local model is completed.
Preferably, the estimating the health parameter of the compressor component based on the method of solving the nonlinear equation and K-means clustering in the step B includes the following steps:
step B2.1, the flow correction coefficient DeltaW of the air compressor is corrected cp As the healthy parameters of the flow of the air compressor, a local model is introduced, and when the rotating speed n is g When no fault occurs, tracking the actual measurement data by the calculation result of the local model, wherein the actual measurement data is directly obtained from a research institute and is obtained by the compressor flow correction coefficient delta W cp Is used as a parameter to construct a training mode nonlinear equation:
in formula (28): the superscript-represents the calculation result of the local model, the Newton Lawson method is adopted to iteratively solve the nonlinear equation, and different conversion rotating speeds n are obtained through calculation g,cor Corresponding compressor flow correction factor DeltaW cp ；
Step B2.2, calculating ΔW due to noise in the signals obtained by the sensor cp Fluctuation exists, and the conversion rotation speed n is converted by means of a K-means clustering method g,cor And correction coefficient DeltaW cp On-line clustering is performed according to Euclidean distance: according to the gas turbine speed n g Judging whether the turboshaft engine is in a steady state or not, generating a new class every time a new quasi-steady state K-means cluster is in, wherein the initial centroid of the class represents the n g,cor Corresponding DeltaW cp Then, the Euclidean distance between the current state and all the centroids is calculated, the current state is classified into the class with the nearest Euclidean distance, and the current state is moreA new centroid; after K-means clustering is finished, linear fitting is carried out to obtain a compressor flow correction coefficient delta W corresponding to each conversion rotating speed cp ＝f(n g,cor )。
The beneficial effects are that:
the invention provides a new method for reconstructing a gas turbine rotating speed signal of a turboshaft engine, and establishes n based on flow path analysis g The local model is used for estimating health parameters by designing a training mode, and compared with the prior art, the method has the following beneficial effects:
1) In comparison with conventional component-level models, n g The internal working mechanism of the turboshaft engine is fully considered by the local model, the effective sensor information is utilized, the number of parts and iteration variables are reduced, the real-time performance is high, and the model can achieve better rotating speed signal reconstruction precision;
2) Study of compressor characteristics versus n g The influence of the calculation precision of the local model is analyzed through a mechanism, the health parameters of the air compressor are decoupled, the training mode is set to estimate the health parameters, and n is improved g Reconstruction accuracy of the local model;
3) The test data verify results show that the method is based on n g The signal reconstruction result of the local model has higher precision in the steady state and transition state processes of the turboshaft engine. Therefore, the method not only can provide analytic redundancy for the fault signal, but also can be used for judging the fault of the hardware sensor in real time. The method uses the existing sensors of the active turboshaft engine, and has good real-time performance, so that the method has higher engineering application value.
Drawings
FIG. 1 is n g Reconstructing a method flow chart;
FIG. 2 is n g A partial model schematic;
FIG. 3 is a schematic view of a bleed air system;
FIG. 4 is a diagram of a training pattern structure;
FIG. 5 shows the different converted rotational speeds n g,cor Corresponding compressor flow correction factor DeltaW cp ；
FIG. 6 shows the compressor flow correction coefficients after clusteringΔW cp ；
FIG. 7 is n based on test data of a civil turboshaft engine g Curve comparison graph;
FIG. 8 is n based on test data of a civil turboshaft engine g Reconstructing an error map of the signal;
Detailed Description
The invention will be further described with reference to the accompanying drawings. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention provides a turbine speed signal reconstruction method of a turboshaft engine based on a local model, which is particularly shown in fig. 1.
Step A, selecting a flow path formed by a compressor to a power turbine inlet guide through flow path analysis of a turboshaft engine, and utilizing the total pressure P of the inlet of the compressor of the active turboshaft engine t2 Total temperature T t2 Total pressure P at compressor outlet t3 And total inlet temperature T of power turbine t45 Sensor for establishing the rotational speed n of the gas turbine of a turboshaft engine g Local model, build n using aerodynamic thermo-mechanical method g And (5) a local model. The modeling object is exemplified by a civil turboshaft engine. The model input is the current gas turbine speed n g (k) And design point rotation speed n g,d Total inlet temperature T of air compressor t2 (k) And the total temperature T of the design point t2,d Total inlet and outlet pressure P of air compressor t2 (k)、P t3 (k) Total temperature T before power turbine t45 (k) Outputting the rotation speed n of the gas turbine at the next moment g (k+1). The effect of combustion delay of the combustion chamber on the engine model is ignored in modeling, and the airflow is regarded as one-dimensional flow, as shown in fig. 2; the method comprises the following steps:
step A1, modeling a bleed air system:
the bleed air system of the modeling object is shown in fig. 3, and the bleed air positions are mainly the bleed air of the middle stage of the air compressor and the bleed air of the outlet, and are used for cooling, sealing and other purposes. Bleed air flow SigmaW j (k) Mainly represented by five parts: compressor intermediate stage to gas turbine inlet guide vane cooling gas W gt1 (k) Compressor intermediate stage to gas turbine outlet cooling gas W gt2 (k) Compressor outlet to power turbine inlet guide vane cooling gas W pt1 (k) The cooling air W from the outlet of the air compressor to the outlet of the power turbine pt2 (k) Air bleed W of other intermediate stages such as air bleed of intermediate stages of air compressor and pneumatic seal los (k) I.e.
∑W j (k)＝W gt1 (k)+W gt2 (k)+W pt1 (k)+W pt2 (k)+W los (k) (1)
The bleed specific enthalpy of the outlet of the air compressor is the specific enthalpy of the outlet airflow of the air compressor, namely
h gt1 (k)＝h gt2 (k)＝h 3 (k) (2)
Intermediate stage bleed specific enthalpy h of compressor cpm (k) Obtained from engineering empirical formula and compressor design data, and the rotation speed n g (k) Enthalpy h of inlet and outlet of compressor 2 (k)、h 3 (k) Related, i.e
h pt1 (k)＝h pt2 (k)＝h los (k)＝h cpm (k)＝f m,h (n g (k),h 2 (k),h 3 (k)) (3)
In the formula, h pt1 (k)、h pt2 (k)、h los (k) The specific enthalpy of the cooling gas at the inlet of the power turbine and the specific enthalpy of the cooling gas at the outlet of the power turbine and the specific enthalpy of the bleed air at the intermediate stage of the compressor are respectively expressed.
Step A2, modeling a compressor:
from current gas turbine speed n g (k) And design rotational speed n g,d Total inlet temperature T of air compressor t2 (k) And the total temperature T of the design point t2,d Total inlet and outlet pressure P of air compressor t2 (k)、P t3 (k) The converted rotating speed n of the compressor can be obtained g,cor (k) Sum pressure ratio pi cp (k)：
Wherein: the subscript d represents the design point parameter and the subscript cor represents the conversion parameter.
By compressor flow characteristics f W,cp And efficiency characteristic f η,cp Can obtain the inlet flow W of the compressor 2 (k) Sum efficiency eta cp (k)：
η cp (k)＝f η,cp (n g,cor (k),π cp (k)) (7)
Calculating the specific enthalpy h of inlet and outlet air flows of the compressor by a variable specific heat method 2 (k)、h 3 (k)
Wherein h, s and f respectively represent the specific enthalpy, specific entropy and oil-gas ratio of the gas, g T2h 、g T2s 、g s2h And g h2T The specific enthalpy is calculated from the total temperature, the specific entropy is calculated from the total temperature, the specific enthalpy is calculated from the specific entropy, and the total temperature is calculated from the specific enthalpy.
Considering bleed air, the compressor outlet flow is:
W 3 (k)＝W 2 (k)-∑W j (k) (9)
the power required by the compressor is as follows:
step A3, modeling a combustion chamber:
in this partial model, the combustor components need only establish power and flow transfer equations, i.e
W 4 (k)＝W 3 (k)+W f (k) (12)
Wherein H is u Is the low heat value eta of the fuel oil b For combustion efficiency
Step A4, turbine modeling:
the turbine model includes flow path calculations from the gas turbine inlet blending to the power turbine inlet blending section. Due to the fact that the total temperature sensor T can pass through the front of the power turbine t45 (k) Obtaining the front specific enthalpy h of the power turbine 45 (k) Only the energy and flow exchange calculations of the bleed air blending section need be performed in the turbine section.
The energy and flow exchange process of the gas turbine inlet guide vane cooling gas blending process may be expressed as:
W 41 (k)＝W 4 (k)+W gt1 (k) (14)
the energy and flow exchange process of the gas turbine outlet cooling gas blend can be expressed as:
W 43 (k)＝W 41 (k) (15)
W 44 (k)＝W 43 (k)+W gt2 (k) (17)
the power turbine inlet guide vane cooling gas blending energy and flow exchange process may be expressed as:
W 45 (k)＝W 44 (k)+W pt1 (k) (19)
the gas turbine emits power as follows:
L gt (k)＝W 41 (k)h 41 (k)-W 43 (k)h 43 (k) (20)
step A5, rotor dynamic model:
from the engine principle, the residual power can be expressed as
ΔL(k)＝η mgt L gt (k)-L cp (k) (21)
Wherein: η (eta) mgt For rotor mechanical efficiency, L gt (k) Generating power for a gas turbine, L cp (k) Absorbing power for the compressor.
After the residual power is calculated by substituting the formulas (1) - (20) into the formula (21), the rotation speed n of the gas turbine at the next moment can be finally calculated g (k+1):
Wherein: ΔL (k) is rotor residual power, T dstep For the simulation step, J is the rotor moment of inertia.
Step B, when n g When the signal does not have a fault, the model enters a training mode, decoupling of health parameters in the local model is realized through theoretical analysis, then the health parameters with the greatest influence on the local model are selected, the health parameters of the component are estimated based on a method for solving a nonlinear equation and K-means clustering according to errors of model calculation results and measured data, and then the health parameters are introduced into n as adjustable parameters g The local model is subjected to online correction, as shown in fig. 4; the method comprises the following steps:
step B1.1, n g The calculation accuracy of (k+1) is very dependent on the remaining power Δl (k). The model is further simplified without consideration of the shaft mechanical efficiency eta mgt And the gas is led out by the gas compressor, the power consumption of the gas compressor can be written as:
L cp (k)＝W 3 (k)h 3 (k)-W 2 (k)h 2 (k) (23)
the gas turbine output power can be written as:
L gt (k)＝W 4 (k)h 4 (k)-W 45 (k)h 45 (k) (24)
the combustor energy exchange can be expressed as:
the remaining power of the core machine is as follows:
ΔL(k)＝L gt (k)-L cp (k) (26)
substituting equations (23) - (25) into equation (26), the remaining power can be written as:
ΔL(k)＝W f (k)H u (k)η b -W f (k)h 45 (k)+W 2 (k)(h 2 (k)-h 45 (k)) (27)
in step B1.2, analysis formula (27) shows that only the compressor flow W affects the residual power DeltaL (k) in the compressor characteristics 2 (k) Without compressor efficiency eta cp . In actual modeling, although the shaft mechanical efficiency η is considered mgt And the bleed air of the compressor due to eta mgt The flow rate of the gas discharged by the gas compressor is close to 1, so that the influence of the efficiency of the gas compressor on the overall calculation accuracy of the model is still small and can be almost ignored. Decoupling of the health parameters of the local model is thereby accomplished.
Step B2.1, when n g When the signal is not invalid, the model output is used for tracking the real engine output, and the compressor flow correction coefficient delta W is used cp Guess values are parameters, and a training mode nonlinear equation is constructed:
in formula (28): superscript-indicates the local model calculation result. Adopts cattleIterative solution of nonlinear equation by using the ton-Lawson method, and calculation to obtain different conversion rotating speeds n g,cor Corresponding compressor flow correction factor DeltaW cp ；
Step B2.2, calculating ΔW due to noise of the sensor cp There is a fluctuation. And carrying out online clustering on the data according to the Euclidean distance by means of a K-means clustering method in pattern recognition. According to the gas turbine speed n g The fluctuation amount of the engine can judge whether the engine is in a steady state, and K-means clustering generates a new class when the engine is in a new quasi-steady state, and the initial centroid of the class represents the n g,cor Corresponding DeltaW cp I.e. (n) g,cor ,ΔW cp ) Then calculate Euclidean distance between the current state and all the current centroidsAnd attributing the current state to the class with the nearest Euclidean distance, and updating the centroid. After K-means clustering is finished, linear fitting is carried out to obtain a compressor flow correction coefficient delta W corresponding to each conversion rotating speed cp ＝f(n g,cor )；
Step C, when the ng signal fails, the model enters a reconstruction mode, and a high-precision ng reconstruction signal is calculated by using the corrected ng local model;
to verify the validity of the gas turbine speed signal reconstruction method, n is verified using certain civil turboshaft engine test data g And selecting four steady-state points and corresponding acceleration and deceleration transition state data in the test run by using the local model. First, the gas turbine rotation speed n in the test run data is set g Total pressure P of inlet of compressor t2 Total temperature T t2 Total pressure P at compressor outlet t3 And total inlet temperature T of power turbine t45 As n g Local model input, model obtains different conversion rotating speeds n through training mode g,cor Corresponding compressor flow correction factor DeltaW cp ＝f(n g,cor ) The results before and after clustering are shown in fig. 5 and 6, and a small number of outliers in the graph are the calculation results of transition state test run data. And introducing the clustered results into the model to complete the online correction of the model.
Re-disconnecting n g Signal input, the model after correction and before correction are operated in the reconstruction mode respectively, n g The signal reconstruction result and test data curve pair is shown in fig. 7, and the error pair is shown in fig. 8. Definition of average steady state error absolute valueT is in 1 For steady state onset time, t 2 For steady-state end time, y (t) is model output, y s And (t) is test run data. Visible n g The local model is basically consistent with the test data curve, the maximum steady-state error before correction is 1.21%, and the average steady-state error of four steady-state points is 1.093%, 1.18%, 1.037% and 1.012% respectively; the maximum steady state error after correction is 0.25%, compared with 79.84% before correction, the average steady state error of four steady state points is 0.0708%, 0.0581%, 0.0495% and 0.0668% respectively, compared with 93.52%, 95.07%, 95.22% and 95.40% before correction; the maximum transition state error before correction is 1.20%, and the average transition state errors of the three transition sections are respectively 0.97%, 1.16% and 1.07%; the maximum transition state error after correction is 0.33%, compared with the maximum transition state error before correction is reduced by 72.5%, the average transition state error of the three transition sections is respectively 0.07%, 0.09% and 0.13%, compared with the maximum transition state error before correction is reduced by 92.78%, 92.24% and 87.85%, and n after correction is visible g The local model can provide n with high accuracy g Reconstructing the signal. The reconstructed rotation speed signal has a certain fluctuation, and a filter can be used for signal processing.
Finally, real-time performance test is carried out, the CPU is operated in a computer with main frequency of 2.6GHz on a MicrosoftVisual C ++ platform, n g The training mode and the reconstruction mode of the local model and the calculation time consumption of the complete component-level model are shown in table 1, and as the number of components of the local model is small and the reconstruction mode is not iterative, the training mode has only 1 iteration variable, the single flow path average calculation time consumption is 0.007ms and 0.035ms respectively, and the calculation time consumption is 4.6% and 23.3% of the average calculation time consumption of the traditional component-level model, and compared with the traditional model, the real-time performance is better.
Table 1 Single step calculation time consuming (ms)
In summary, the new method for reconstructing the gas turbine rotational speed signal of the turboshaft engine provided by the invention selects the existing partial effective sensor information and components of the turboshaft engine, and establishes a gas turbine rotational speed partial model. The health parameters in the local model are decoupled and estimated on line through mechanism analysis, so that the local model is corrected on line. The test data verification of a civil turboshaft engine shows that the steady state error of the reconstruction signal is not more than 0.25 percent and the transition state error is not more than 0.33 percent, and the signal reconstruction method has high precision. The method does not need to add an additional sensor, and has good real-time performance and strong engineering practicability.
The foregoing is only a preferred embodiment of the invention, it being noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the invention.

Claims (4)

1. The method for reconstructing the gas turbine rotating speed signal of the turboshaft engine based on the local model is characterized by comprising the following steps of:
step A, establishing the rotation speed n of the gas turbine g A local model;
step B, when the rotating speed n g When no fault occurs, the gas turbine speed n g The local model enters a training mode, decoupling of two health parameters, namely the flow and the efficiency of the gas compressor in the local model, is realized through theoretical analysis, then the health parameter with the greatest influence on the local model is selected, the health parameter of the gas compressor component is estimated based on a method for solving a nonlinear equation and K-means clustering according to the error between the calculation result and the measured data of the local model, and then the health parameter is used as an adjustable parameter to be introduced into the local model, so that the online correction of the local model is completed;
step C, when n g When the signal fails, the local model enters a reconstruction mode, and the corrected local model is used for calculating high-precision n g Reconstructing the signal.
2. The method for reconstructing a gas turbine rotational speed signal of a turboshaft engine based on a local model of claim 1, wherein the gas turbine rotational speed local model in step a is constructed as follows:
step A1, bleed air system Sigma W j (k) Modeling:
bleed air flow Sigma W j (k) Comprising the following steps:
intermediate stage of compressor to gas turbine inlet cooling gas W gt1 (k) Compressor intermediate stage to gas turbine outlet cooling gas W gt2 (k) Cooling gas W from outlet of gas compressor to inlet of power turbine pt1 (k) The cooling air W from the outlet of the air compressor to the outlet of the power turbine pt2 (k) Other intermediate stage bleed air W los (k) The other intermediate stage bleed air comprises air bleed air of an intermediate stage of a compressor, and the bleed air is sealed pneumatically:
∑W j (k)＝W gt1 (k)+W gt2 (k)+W pt1 (k)+W pt2 (k)+W los (k) (1)
step A2, modeling a compressor:
at time k, bleed air specific enthalpy h of outlet of compressor gt1 (k) Specific enthalpy of inlet air flow h of air compressor 2 (k) Specific enthalpy of compressor outlet flow h 3 (k) The following relationship exists:
h gt1 (k)＝h gt2 (k)＝h 3 (k) (2)
intermediate stage bleed air specific enthalpy h of compressor at moment k cpm (k) And gas turbine speed n g (k) Specific enthalpy of inlet air flow h of air compressor 2 (k) Specific enthalpy of compressor outlet flow h 3 (k) The method meets the following conditions:
h pt1 (k)＝h pt2 (k)＝h los (k)＝h cpm (k)＝f m,h (n g (k),h 2 (k),h 3 (k)) (3)
in the formula, h pt1 (k) Representing powerSpecific enthalpy of turbine inlet cooling gas, h pt2 (k) Represents the specific enthalpy of cooling gas at the outlet of the power turbine, h los (k) Representing the specific enthalpy of other intermediate stage bleed air; h is a cpm (k) By the gas turbine speed n at time k g (k) Specific enthalpy h of inlet air flow of compressor 2 (k) And compressor outlet air flow specific enthalpy h 3 (k) Calculated, the process uses f m,h A function representation;
according to the gas turbine rotation speed n at time k g (k) Design speed n of gas turbine g,d Total inlet temperature T of air compressor t2 (k) Total temperature T of inlet design point of air compressor t2,d Total pressure P of inlet of compressor t2 (k) And the total pressure P of the outlet of the air compressor t3 (k) Obtaining the converted rotating speed n of the air compressor g,cor (k) Sum pressure ratio pi cp (k)：
Wherein: the subscript d represents the design point parameter of the inlet of the compressor, and the subscript cor represents the conversion parameter;
by compressor flow characteristics f W,cp Obtaining the inlet flow W of the air compressor 2 (k) By compressor efficiency characteristics f η,cp Obtaining the efficiency eta of the compressor cp (k)：
η cp (k)＝f η,cp (n g,cor (k),π cp (k)) (7)
Calculating the specific enthalpy h of the inlet air flow of the air compressor by a fixed specific heat method 2 (k) Specific enthalpy of compressor outlet flow h 3 (k)：
Wherein, h, s and f respectively represent specific enthalpy, specific entropy and oil-gas ratio, g T2h 、g T2s 、g s2h And g h2T As a function, the corner mark 2 represents the compressor inlet, 3 represents the compressor outlet, i represents the flow loss of gas, T t3 (k) The total inlet temperature of the air compressor;
considering the bleed air of the compressor, the outlet flow of the compressor is as follows:
W 3 (k)＝W 2 (k)-∑W j (k) (9)
the power required by the compressor is as follows:
step A3, modeling a combustion chamber:
the combustor components need only establish energy and flow transfer equations, where the energy transfer equations are:
the flow transfer equation the transfer equation is:
W 4 (k)＝W 3 (k)+W f (k) (12)
wherein H is u Is the low heat value eta of the fuel oil b For combustion efficiency, 4 denotes the combustion chamber outlet cross section, h 4 (k) Indicating the specific enthalpy of the outlet airflow of the combustion chamber at the moment k, W f (k) The fuel flow of the combustion chamber at the moment k is represented;
step A4, turbine modeling:
the energy exchange process of the gas turbine inlet cooling gas blending process is represented as:
the flow exchange process of the gas turbine inlet cooling gas blending process is represented as:
W 41 (k)＝W 4 (k)+W gt1 (k) (14)
in the formula, h 41 (k) Represents the section airflow specific enthalpy after mixing the inlet of the gas turbine at the moment k, W 41 (k) Representing the cross-sectional gas flow rate after blending at the gas turbine inlet; and has
W 43 (k)＝W 41 (k) (15)
The energy exchange process of the gas turbine outlet cooling gas blending process is represented as:
the flow exchange process of the gas turbine outlet cooling gas blending process is represented as:
W 44 (k)＝W 43 (k)+W gt2 (k) (17)
in which W is 43 (k) Represents the gas flow rate of the cross section before blending of the gas turbine outlet, h 43 (k) Representing the specific enthalpy of the section airflow before mixing of the outlet of the gas turbine at the moment k, W 44 (k) Representing the gas flow rate after mixing at the outlet of the gas turbine at the time k;
the power turbine inlet cooling gas blending energy exchange process is expressed as:
the power turbine inlet cooling gas blending flow rate exchange process is expressed as:
W 45 (k)＝W 44 (k)+W pt1 (k) (19)
in the formula, h 45 (k) Represents the specific enthalpy of the section airflow before blending of the inlet of the power turbine at the moment k, W 45 (k) The gas flow rate after mixing at the inlet of the power turbine at the moment k is shown;
the gas turbine emits power as follows:
L gt (k)＝W 41 (k)h 41 (k)-W 43 (k)h 43 (k) (20)
step A5, rotor dynamics model:
the residual power of the core machine is expressed as follows according to the principle of the engine
ΔL(k)＝η mgt L gt (k)-L cp (k) (21)
Wherein: η (eta) mgt For rotor mechanical efficiency, L gt (k) Generating power for a gas turbine, L cp (k) The power required by the compressor;
substituting equations (1) - (20) into equation (21) to calculate the residual power, and finally calculating the next-time rotational speed n of the gas turbine g (k+1):
Wherein: ΔL (k) is rotor residual power, T dstep For the simulation step, J is the rotor moment of inertia.
The local model comprises a constructed air entraining system Sigma W i (k) Compressor, combustor, turbine, and rotor dynamics model.
3. The method for reconstructing the gas turbine rotational speed signal of the turboshaft engine based on the local model according to claim 2, wherein the decoupling process of the two health parameters of the flow and the efficiency of the compressor in the local model is realized through theoretical analysis in the step B as follows:
step B1.1, since the decoupling process does not require the construction of a complex local model in step A, i.e. the shaft mechanical efficiency η is not taken into account mgt And the gas is discharged by the gas compressor, and then the local model is simplified, so that the power required by the gas compressor is as follows:
L cp (k)＝W 3 (k)h 3 (k)-W 2 (k)h 2 (k) (23)
the gas turbine emits power as follows:
L gt (k)＝W 4 (k)h 4 (k)-W 45 (k)h 45 (k) (24)
the combustor energy exchange is expressed as:
the remaining power of the core machine is as follows:
ΔL(k)＝L gt (k)-L cp (k) (26)
substituting equations (23) - (25) into equation (26), the core residual power is written as:
ΔL(k)＝W f (k)H u (k)η b -W f (k)h 45 (k)+W 2 (k)(h 2 (k)-h 45 (k)) (27)
in step B1.2, analysis (27) shows that the only healthy parameter affecting the residual power ΔL (k) in the compressor characteristics is the compressor flow W 2 (k) In the modeling of step A, the mechanical efficiency η of the shaft is taken into account mgt And the bleed air of the compressor due to eta mgt And the flow of the gas discharged by the gas compressor is ignored and is close to 1, so that the influence of the efficiency of the gas compressor on the whole calculation accuracy of the local model is small and is ignored, and the decoupling of the health parameters of the local model is completed.
4. The method for reconstructing the rotational speed signal of the gas turbine of the turboshaft engine based on the local model as set forth in claim 3, wherein the estimating the health parameters of the compressor component based on the method for solving the nonlinear equation and the K-means clustering in the step B comprises the following steps:
step B2.1, the flow correction coefficient DeltaW of the air compressor is corrected cp As the healthy parameters of the flow of the air compressor, a local model is introduced, and when the rotating speed n is g When no fault occurs, tracking the actual measurement data by the calculation result of the local model, wherein the actual measurement data is directly obtained from a research institute and is obtained by the compressor flow correction coefficient delta W cp Is used as a parameter to construct a training mode nonlinear equation:
in formula (28): the superscript-represents the calculation result of the local model, the Newton Lawson method is adopted to iteratively solve the nonlinear equation, and different conversion rotating speeds n are obtained through calculation g,cor Corresponding compressor flow correction factor DeltaW cp ；
Step B2.2, calculating ΔW due to noise in the signals obtained by the sensor cp Fluctuation exists, and the conversion rotation speed n is converted by means of a K-means clustering method g,cor And correction coefficient DeltaW cp On-line clustering is performed according to Euclidean distance: according to the gas turbine speed n g Judging whether the turboshaft engine is in a steady state or not, generating a new class every time a new quasi-steady state K-means cluster is in, wherein the initial centroid of the class represents the n g,cor Corresponding DeltaW cp Then calculating Euclidean distances between the current state and all the current centroids, attributing the current state to the class with the nearest Euclidean distance, and updating the centroids; after K-means clustering is finished, linear fitting is carried out to obtain a compressor flow correction coefficient delta W corresponding to each conversion rotating speed cp ＝f(n g,cor )。
CN202311493674.5A 2023-11-10 2023-11-10 Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model Pending CN117521506A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311493674.5A CN117521506A (en) 2023-11-10 2023-11-10 Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311493674.5A CN117521506A (en) 2023-11-10 2023-11-10 Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model

Publications (1)

Publication Number Publication Date
CN117521506A true CN117521506A (en) 2024-02-06

Family

ID=89743227

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311493674.5A Pending CN117521506A (en) 2023-11-10 2023-11-10 Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model

Country Status (1)

Country Link
CN (1) CN117521506A (en)

Similar Documents

Publication Publication Date Title
CN111914362B (en) Self-adaptive method for turbofan engine model in research and development stage
Zamboni et al. Fan root aerodynamics for large bypass gas turbine engines: Influence on the engine performance and 3D design
CN106874569B (en) Variable-geometry split-shaft type gas turbine real-time simulation modeling method
CN108829928B (en) Turboshaft engine adaptive component-level simulation model construction method
US8862433B2 (en) Partitioning of turbomachine faults
CN108108528B (en) One-dimensional matching design method for power turbine of split-shaft type aeroderivative
CN108733906B (en) Method for constructing aero-engine component level model based on accurate partial derivative
CN110502840A (en) Aerial engine air passage parameter on-line prediction method
CN106126902A (en) Gas turbine component characteristic line modification method based on particle swarm optimization algorithm identification
CN109871653B (en) Method for correcting part characteristics of aircraft engine mathematical model
CN105912878A (en) Gas turbine adaptive gas circuit component performance diagnostic method based on combination of thermal model and particle swarm optimization
CN114154234A (en) Modeling method, system and storage medium for aircraft engine
CN116663202B (en) Checking method of performance simulation model of multistage axial flow compressor
JP5845705B2 (en) Gas turbine performance estimation device
CN117521506A (en) Vortex shaft engine gas turbine rotating speed signal reconstruction method based on local model
Gundy-Burlet et al. Influence of 3D hot streaks on turbine heat transfer
Dahlquist et al. The influence from the working medium on the profile loss in compressor and turbine airfoils
CN114510798A (en) Pneumatic axial force error-proofing analysis method for aircraft engine
Kim et al. Suitability of performance adaptation methods for updating the thermodynamic cycle model of a turboprop engine
Yamagami et al. Unsteady flow effects in a high-speed multistage axial compressor
CN111611743A (en) Axial-flow compressor characteristic line self-adaption method
CN114818205B (en) Online sensing method for blade tip clearance of full life cycle of aero-engine
US20220236701A1 (en) Method for determining a predictive model of a pressure ratio for a dual-flow turbine engine
Berdanier et al. Scaling Sealing Effectiveness in a Stator-Rotor Cavity for Differing Blade Spans
Purushothaman et al. Cold blade profile generation methodology for compressor rotor blades using FSI approach

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination