WO2020118512A1 - 一种基于lft的航空发动机传感器及执行机构故障诊断方法 - Google Patents

一种基于lft的航空发动机传感器及执行机构故障诊断方法 Download PDF

Info

Publication number
WO2020118512A1
WO2020118512A1 PCT/CN2018/120243 CN2018120243W WO2020118512A1 WO 2020118512 A1 WO2020118512 A1 WO 2020118512A1 CN 2018120243 W CN2018120243 W CN 2018120243W WO 2020118512 A1 WO2020118512 A1 WO 2020118512A1
Authority
WO
WIPO (PCT)
Prior art keywords
matrix
fault
aero
engine
variable
Prior art date
Application number
PCT/CN2018/120243
Other languages
English (en)
French (fr)
Inventor
马艳华
杜宪
汪锐
刘敏
孙希明
Original Assignee
大连理工大学
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 大连理工大学 filed Critical 大连理工大学
Priority to US16/604,042 priority Critical patent/US11203446B2/en
Priority to PCT/CN2018/120243 priority patent/WO2020118512A1/zh
Publication of WO2020118512A1 publication Critical patent/WO2020118512A1/zh

Links

Images

Classifications

    • BPERFORMING OPERATIONS; TRANSPORTING
    • B64AIRCRAFT; AVIATION; COSMONAUTICS
    • B64FGROUND OR AIRCRAFT-CARRIER-DECK INSTALLATIONS SPECIALLY ADAPTED FOR USE IN CONNECTION WITH AIRCRAFT; DESIGNING, MANUFACTURING, ASSEMBLING, CLEANING, MAINTAINING OR REPAIRING AIRCRAFT, NOT OTHERWISE PROVIDED FOR; HANDLING, TRANSPORTING, TESTING OR INSPECTING AIRCRAFT COMPONENTS, NOT OTHERWISE PROVIDED FOR
    • B64F5/00Designing, manufacturing, assembling, cleaning, maintaining or repairing aircraft, not otherwise provided for; Handling, transporting, testing or inspecting aircraft components, not otherwise provided for
    • B64F5/60Testing or inspecting aircraft components or systems
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00Testing or monitoring of control systems or parts thereof
    • G05B23/02Electric testing or monitoring
    • G05B23/0205Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07CTIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
    • G07C5/00Registering or indicating the working of vehicles
    • G07C5/08Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
    • G07C5/0808Diagnosing performance data
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/30Nc systems
    • G05B2219/45Nc applications
    • G05B2219/45071Aircraft, airplane, ship cleaning manipulator, paint stripping

Definitions

  • the invention belongs to the field of aircraft engine fault diagnosis, and in particular relates to a LFT-based aircraft engine sensor and actuator failure diagnosis method.
  • An aero engine is an important part of an aircraft.
  • the health of the flight state depends directly on the health of the aero engine.
  • the sensor of the aero engine is used as the bottom information collection platform of the aero engine, which can accurately measure the state information of the components and the system in the process of the aero engine, thereby constructing an effective control system. If the aero-engine sensor fails, it is impossible to provide accurate performance and status parameters to the control system, and accurate control cannot be achieved.
  • the aero-engine actuator is an important link between the aero-engine and the control system. According to the control instructions given by the control system, the actuator drives the actual control variable to control the working state of the aero-engine.
  • the present invention monitors the performance of aero-engine sensors and actuators and has real-time diagnosis and warning of fault conditions, which is of great significance.
  • the existing literature shows that, first of all, the research direction of the fault diagnosis technology of the existing aero-engine sensors and actuators is mainly focused on the fault detection, that is, only the aero-engine sensor and the actuators can be judged whether the fault occurs, and the fault There are few studies on signal estimation methods. Specifically, different fault modes of sensors and actuators correspond to different fault handling measures. For the sensor failure of the aero engine, the actual failure is mainly drift. If the sensor drifts, the sensor measurement information can be corrected by the controller design. The failure of the aero engine actuator includes degradation, drift and stuck. If the actuator is degraded or drifted, the controller can be designed to keep the aero engine working normally; if the actuator is stuck, the actuator needs to be switched. To the redundant mode, perform actuator maintenance after the flight is completed.
  • the aero engine can be described as a typical LPV (linear variable parameter) system.
  • the literature shows that the fault estimation methods of LPV systems have been divided into two major categories in recent years: one is the observer-based method, but this type of method is not robust to the disturbance in the system and the model uncertainty, that is, the system is subjected to Errors caused by external disturbances and modeling will seriously affect the observation results of the observer; the other is the fault estimation method based on H ⁇ optimization technology.
  • This method can improve the robustness of the system, but the fault estimation research based on this technology It is still in its infancy, and there are still many issues worthy of in-depth discussion.
  • the present invention provides an LFT-based aero-engine sensor and actuator fault diagnosis method, which can According to the changes in the parameters of the aeroengine LPV model, adaptively adjust the parameters of the fault estimator to achieve rapid detection of faults in the system, and accurately reconstruct the fault signal, promptly propose repairs, and provide better control for subsequent fault tolerance The basic guarantee.
  • An LFT-based aero-engine sensor and actuator fault diagnosis method includes the following steps:
  • Step 1 The method of combining small disturbance method and linear fitting method to establish aero-engine state variable model
  • Step 2 Establish a linear variable parameter (LPV) model of the aero engine dependent on affine parameters
  • Step 3 Transform the aeroengine LPV model dependent on the affine parameters of disturbances and sensor and actuator failures into an LFT structure, and establish an H ⁇ integrated framework of the aeroengine LPV fault estimator;
  • Step 4 Solve a set of linear matrix inequalities LMIs to obtain the solution condition of the fault estimator
  • Step 5 Design the fault estimator in combination with the LFT structure to realize the fault diagnosis of aero-engine sensors and actuators.
  • the step 1 includes the following steps:
  • Step 1.1 Input the oil pressure p f at the steady-state operating point to the aero engine, and after the relative conversion speed n h of the aero engine high pressure turbine reaches the corresponding stable state, input the oil pressure step signal U pf1 with an amplitude of 0.01 p f To the aero engine, collect the aero engine output high-pressure turbine relative conversion speed response Y nh1 and low-pressure turbine relative conversion speed response Y nl1 respectively ;
  • Step 1.3 Taking the oil pressure step signal U pfi as the input variable, the high-pressure turbine relative conversion speed response Y nhi and the low-pressure turbine relative conversion speed response Y nli as the state variables, the aero-engine under each steady-state operating point is solved according to the linear fitting method Discrete small deviation state variable model;
  • Step 1.4 According to the sampling period T, convert the discrete small deviation state variable model of the aero engine at each steady-state operating point into a continuous small deviation state variable model to obtain the aero engine state variable model;
  • the step 2 includes the following steps:
  • Step 2.2 Express the system matrix A p ( ⁇ ) and control matrix B p ( ⁇ ) of the aero engine continuous small deviation state variable model as affine parameter dependent forms, as follows:
  • a 0 , A 1 , B 0 , and B 1 respectively represent coefficient matrices to be sought.
  • I is the identity matrix
  • [I ⁇ I] + is the Moore-Penrose pseudo-inverse of [I ⁇ I], that is, the system matrix A p ( ⁇ ) and the control matrix B p ( ⁇ ) of the aeroengine LPV model dependent on the affine parameters are obtained;
  • Step 2.3 Establish LPV model of aero-engines dependent on affine parameters
  • step 3 establishes the H ⁇ synthesis framework of the aeroengine LPV fault estimator including the following steps:
  • Step 3.1 Represent the aero-engine LPV model P(s, ⁇ ) with disturbances and sensor and actuator failure affine parameter dependence as
  • d ⁇ R q is the disturbance signal
  • f ⁇ R l fault signals including actuators and sensor failure fault
  • R q, R l each dimension is represented by q, l the set of real numbers
  • E p, F p, G p and H p are system state space matrices
  • the upper LFT structure of P(s, ⁇ ) is expressed as
  • Step 3.2 Let the fault estimator K(s, ⁇ ) be of the form
  • a K , B K1 , B K ⁇ , C K1 , C K ⁇ , D K11 , D K1 ⁇ , D K ⁇ 1 , D K ⁇ are the system state space matrix
  • Step 3.3 change portion ⁇ ( ⁇ ) according to the time aeroengine LPV model P (s, ⁇ ) and fault estimator K (s, ⁇ ) when the change portion ⁇ K ( ⁇ ), H ⁇ LPV fault estimator Comprehensive
  • the framework is expressed as
  • the step 4 obtains the solvable condition for the existence of the fault estimator including the following steps:
  • Step 4.1 Obtain the solution condition for the existence of the fault estimator K(s, ⁇ ), namely
  • X is positive definite symmetric matrix, full block scalar matrix It is a symmetric matrix, and ⁇ >0 is the performance index; Q, S, and R represent the sub-scalar matrix blocks of P, respectively.
  • Step 4.2 Block the positive definite symmetric matrix X and its inverse matrix X -1
  • L, M, E respectively represent the matrix block of X
  • J, N respectively represent the sub-matrix block of X -1 .
  • Q 1 , Q 2 and Q 3 respectively represent the sub-matrix blocks of Q
  • S 1 , S 2 , S 3 and S 4 respectively represent the sub-matrix blocks of S
  • R 1 , R 2 and R 3 respectively represent the sub-sub-blocks of R Matrix block, Respectively Sub-matrix block of, Respectively Sub-matrix block of, Respectively Sub-matrix block of, Respectively Of sub-matrix blocks.
  • N L and N J represent [C 2 D 2 ⁇ D 21 ] and The base of nuclear space;
  • Step 4.3 Solve linear matrix inequalities (15)-(18) to obtain matrix solutions L, J, Q 3 , S 4 ,
  • the step 5 combined with the LFT structural design fault estimator includes the following steps:
  • Step 5.1 According to the obtained matrix solution L, J, Q 3 , S 4 , Find the positive definite symmetric matrix X, the full block scalar matrix P and its inverse matrix from equations (13) and (14)
  • Step 5.2 According to Schur's complement theorem, express the linear matrix inequality (11) as
  • Step 5.3 Obtain the state space matrix of the fault estimator K(s, ⁇ )
  • the beneficial effects of the present invention are: through the aero-engine sensor and actuator fault diagnosis method designed by the present invention, the aero-engine LPV model and the fault estimator are converted into a time-invariant part and a time-varying part LFT structure, in which the time-varying The part changes with the change of the time-varying parameter vector, so that the fault estimator has gain scheduling characteristics, which can realize the accurate estimation of the fault signal under the influence of uncertain conditions such as external disturbances and modeling errors, so that it is convenient to understand the type of fault, Generate information such as time and severity.
  • the present invention reduces the conservativeness of the fault estimator design through the S process.
  • Figure 3 shows the upper LFT structure of the aeroengine LPV model P(s, ⁇ ).
  • Figure 4 is a system structure diagram under the LFT framework.
  • Figure 5 shows the H ⁇ synthesis framework of the LPV fault estimator.
  • Figure 6(a) and Figure 6(b) are the simulation results of catastrophic failure estimation.
  • Fig. 7(a) and Fig. 7(b) are the simulation results of the slowly varying fault estimation.
  • Figures 8(a) and 8(b) are the simulation results of intermittent fault estimation.
  • FIG. 9 is a schematic flowchart of the present invention.
  • Step 1.1 Input the oil pressure p f at the steady-state operating point to the aero engine, and after the relative conversion speed n h of the aero engine high pressure turbine reaches the corresponding stable state, input the oil pressure step signal U pf1 with an amplitude of 0.01 p f To the aero-engine, collect the aero-engine output high-pressure turbine relative conversion speed response Y nh1 and low-pressure turbine relative conversion speed response Y nl1 respectively .
  • fi high pressure turbine in response to the relative rotational speed converted Y nhi
  • Step 1.3 Taking the oil pressure step signal U pfi as the input variable, the high-pressure turbine relative conversion speed response Y nhi and the low-pressure turbine relative conversion speed response Y nli as the state variables, then the aerospace engine discrete small deviation state variable model is expressed as
  • the state variable x p [Y nl Y nh ] T ⁇ R n
  • the input variable u U pf ⁇ R t
  • the output variable y p Y nh ⁇ R m
  • i 1, 2, 3, ..., 13
  • Subscripts k, k+1 are the corresponding sampling times
  • E i , F i , G i , H i are the system space matrix of the appropriate dimension
  • R n , R t , R m represent the dimension n, respectively Set of real numbers of t and m
  • T means transpose the matrix.
  • Step 2.2 Express the system matrix A p ( ⁇ ) and the control matrix B p ( ⁇ ) of the aero engine continuous small deviation state variable model system as affine parameter dependent forms, as follows:
  • a 0 , A 1 , B 0 , and B 1 respectively represent the coefficient matrix to be sought.
  • I is the identity matrix
  • Step 2.3 Establish LPV model of aero-engines dependent on affine parameters
  • Step 3.1 Represent the aero-engine LPV model P(s, ⁇ ) with disturbances and sensor and actuator failure affine parameter dependence as
  • d ⁇ R q is the disturbance signal
  • the Gaussian white noise with a standard deviation of 0.001 is taken
  • f ⁇ R l is the fault signal, including sensor fault and actuator fault
  • the mutation fault, slow-change fault and intermittent fault are taken respectively
  • R q , R l represents the set of real numbers with dimensions q and l respectively
  • G p 0.2
  • H p 1.
  • Fu represents the upper LFT structure
  • P′ represents the time-invariant part of P(s, ⁇ )
  • the sum of the dimension t of u, the dimension q of the disturbance signal d and the dimension l of the fault signal f; the system state space matrix is
  • Step 3.2 Let the fault estimator K(s, ⁇ ) be of the form
  • xK ⁇ R k is the state variable of the fault estimator K(s, ⁇ )
  • a K ( ⁇ ), B K ( ⁇ ), C K ( ⁇ ), D K ( ⁇ ) is the system state space matrix.
  • K(s, ⁇ ) as the following LFT structure, as follows:
  • K′ represents the time-invariant part of K(s, ⁇ )
  • a K , B K1 , B K ⁇ , C K1 , C K ⁇ , D K11 , D K1 ⁇ , D K ⁇ 1 , and D K ⁇ are system state space matrices of appropriate dimensions.
  • Step 3.3 The system connection diagram under the LFT framework is shown in Figure 4, then the state space expression of system P 1 in Figure 4 is
  • system matrix A A p
  • system matrix B ⁇ B p ⁇
  • system matrices B 1 B pw
  • the system matrix B 2 0 n ⁇ l
  • the system matrix C ⁇ C p ⁇
  • the system matrix D ⁇ D p ⁇
  • the system matrix D ⁇ 1 D p ⁇ w
  • the system matrix D ⁇ 2 0 r ⁇ l
  • the system matrix C 1 0 p1 ⁇ n
  • the system matrix D 1 ⁇ 0 p1 ⁇ r
  • system matrix System matrix System matrix System matrix System matrix System matrix System matrix System matrix D 22 0 p2 ⁇ l
  • n represents the dimension of the state variable x p of the aeroengine
  • r represents the output variable w ⁇ of the time-varying part ⁇ ( ⁇ ) and the output variable w K of the time-varying part ⁇ K ( ⁇ ) dimension.
  • H ⁇ integrated framework LPV fault estimator may represent The following formula, as shown in Figure 5
  • Is the fault estimation error that is, the output variable of the H ⁇ synthesis frame of the LPV fault estimator
  • the system matrix System matrix System matrix System matrix System matrix System matrix System matrix System matrix System matrix Fault estimator matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix matrix
  • Step 4.1 If there is a symmetric positive definite matrix X, symmetric matrix So that equation (37) and equation (38) are established,
  • the closed-loop system (36) is asymptotically stable, and the L 2 induced norm of the closed-loop transfer function from the external input w to the fault estimation error e f is less than the performance index ⁇ ( ⁇ >0). That is, the solvable conditions for the existence of the fault estimator K(s, ⁇ ) are Equation (37) and Equation (38). Among them, Q, S, R respectively represent the sub-matrix block of P.
  • Step 4.2 Block the positive definite symmetric matrix X and its inverse matrix X -1
  • L, M, E respectively represent the matrix block of X
  • J, N respectively represent the sub-matrix block of X -1 .
  • Q 1 , Q 2 and Q 3 respectively represent the sub-matrix blocks of Q
  • S 1 , S 2 , S 3 and S 4 respectively represent the sub-matrix blocks of S
  • R 1 , R 2 and R 3 respectively represent the sub-sub-blocks of R Matrix block, Respectively Sub-matrix block of, Respectively Sub-matrix block of, Respectively Sub-matrix block of, Respectively Of sub-matrix blocks.
  • the linear matrix inequality (37) is organized as
  • Step 4.3 The linear matrix inequality (42) is organized as
  • U ⁇ and V ⁇ are the basis of the nuclear space of U T and V, respectively.
  • N L and N J represent [C 2 D 2 ⁇ D 21 ] and The base of the nuclear space.
  • Step 4.4 Use the LMI toolbox in MATLAB to solve the linear matrix inequalities (40), (44), (47), (48), and obtain the optimal ⁇ value of 0.21, and the corresponding matrix solutions L, J, Q 3 . S 4 ,
  • Step 5.1 According to the obtained matrix solution L, J, Q 3 , S 4 , Find the positive definite symmetric matrix X, the full block scalar matrix P and its inverse matrix from equations (39) and (41)
  • Step 5.2 According to Schur complement theorem, the linear matrix inequality (37) is expressed as
  • Step 5.3 Obtain the state space matrix of the fault estimator K(s, ⁇ )
  • the simulation results show that the fixed parameter fault estimator designed by the standard H ⁇ method cannot cope well with the change of variable parameters, and the LPV fault estimator designed by the present invention can quickly detect the fault in the system and accurately reconstruct the fault signal, Has obvious performance advantages.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Manufacturing & Machinery (AREA)
  • Transportation (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)
  • Feedback Control In General (AREA)

Abstract

一种基于LFT的航空发动机传感器及执行机构故障诊断方法,包括以下步骤:采用小扰动法与线性拟合法相结合的方法建立航空发动机状态变量模型,并基于此模型建立仿射参数依赖的航空发动机LPV模型;将带有扰动信号及传感器与执行机构故障信号的航空发动机LPV模型转化为LFT结构,得到LPV故障估计器的H ∞综合框架;通过求解一组线性矩阵不等式LMIs获得故障估计器存在的有解条件;结合LFT结构设计故障估计器,实现航空发动机传感器及执行机构的故障诊断。所述诊断方法根据航空发动机参数的变化,自适应地调整故障估计器的参数,迅速地检测传感器及执行机构故障,准确重构故障信号,为后续的主动容错控制提供基础。

Description

一种基于LFT的航空发动机传感器及执行机构故障诊断方法 技术领域
本发明属于航空发动机故障诊断领域,具体涉及一种基于LFT的航空发动机传感器及执行机构故障诊断方法。
背景技术
航空发动机是飞机的重要部件,飞行状态健康与否直接取决于航空发动机的健康状态。其中,航空发动机的传感器作为航空发动机的底层信息采集平台,能够准确测量航空发动机工作过程中的部件和系统状态信息,从而构建出有效的控制系统。若航空发动机传感器发生故障,则无法给控制系统提供精确的性能和状态参数,无法实现精确的控制。而航空发动机的执行机构是连接航空发动机和控制系统的重要环节,执行机构根据控制系统给定的控制指令,驱动实际控制量变化,从而控制航空发动机的工作状态。若航空发动机执行机构发生故障,则会给控制系统提供错误的信息,从而给飞机的安全带来隐患,有可能引发灾难性的后果。因此,本发明针对航空发动机传感器及执行机构的性能进行监测,对故障状态实时诊断、告警,具有重要的意义。
已有文献表明,首先,现有的航空发动机传感器及执行机构的故障诊断技术的研究方向主要集中在故障检测方面,即只能对航空发动机传感器及执行机构是否发生故障给出判断,而针对故障信号进行估计方法的研究较少。具体的,传感器及执行机构不同的故障模式对应的故障处理措施不同。对于航空发动机的传感器故障,实际故障以漂移为主,若传感器发生漂移,则可通过控制器设计对传感器测量信息进行校正。航空发动机的执行机构故障包括退化、漂移与卡死等状态,若执行机构发生退化或漂移,则可通过控制器设计使航空发动机保持正常工作;若执行机构发生卡死,则需要进行执行机构切换到冗余模式,待飞行完成后进行执行机构维护。因此,只对故障检测进行研究,即判断是否发生故障,不利于航空发动机的安全运行与维护。而对故障程度可靠的估计,可实现航空发动机传感器及执行机构状态的准确判断,同时降低航空发动机的维修成本。此外,航空发动机可以描述为一种典型的LPV(线性变参数)系统。文献表明,LPV系统的故障估计方法近年来主要分为两大类:一是基于观测器的方法,但是这类方法对系统中的扰动以及模型不确定性鲁棒性不强,即系统受到的外界扰动和建模带来的误差会严重影响观测器的观测结果;另一类是基于H 优化技术的故障估计方法,该方法可以提高系统的鲁棒性,但基于该技术的故障估计研究仍处于刚刚起步的阶段,还有许多问题值得深入地探讨。
发明内容
针对现有航空发动机传感器及执行机构故障诊断技术中无法实现外界干扰及建模误差下对故障信号进行准确估计的问题,本发明提供一种基于LFT的航空发动机传感器及执行机构故障诊断方法,能够根据航空发动机LPV模型中参数的变化,自适应地调整故障估计器的参数,实现迅速地检测系统中的故障,并准确地重构故障信号,及时提出维修建议,为之后的容错控制提供更好的基础保障。
为实现上述目的,本发明采用的技术方案步骤如下:
一种基于LFT的航空发动机传感器及执行机构故障诊断方法,包括以下步骤:
步骤1:采用小扰动法与线性拟合法相结合的方法建立航空发动机状态变量模型;
步骤2:建立仿射参数依赖的航空发动机线性变参数(LPV)模型;
步骤3:将存在扰动及传感器与执行机构故障的仿射参数依赖的航空发动机LPV模型转化为LFT结构,建立航空发动机LPV故障估计器的H 综合框架;
步骤4:求解一组线性矩阵不等式LMIs,获得故障估计器存在的有解条件;
步骤5:结合LFT结构设计故障估计器,实现航空发动机传感器及执行机构的故障诊断。
所述步骤1包括以下步骤:
步骤1.1:向航空发动机输入稳态工作点下的油压p f,待航空发动机高压涡轮相对换算转速n h达到相应稳定状态后,将幅值为0.01p f的油压阶跃信号U pf1输入到航空发动机中,分别采集航空发动机输出高压涡轮相对换算转速响应Y nh1和低压涡轮相对换算转速响应Y nl1
步骤1.2:重复步骤1的过程N次,分别采集得到给定油压p fi下的高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli,i=1,2,3,…,N;
步骤1.3:以油压阶跃信号U pfi为输入变量,高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli为状态变量,根据线性拟合法求解各稳态工作点下的航空发动机离散小偏差状态变量模型;
步骤1.4:根据采样周期T,将各稳态工作点下的航空发动机离散小偏差状态变量模型转换为连续小偏差状态变量模型,得到航空发动机状态变量模型;
Figure PCTCN2018120243-appb-000001
其中,状态变量x p=[Y nl T Y nh T] T∈R n
Figure PCTCN2018120243-appb-000002
表示x p的一阶导数,输入变量u=U pf∈R t,输出变量y p=Y nh∈R m,A pi、B pi、C pi、D pi是系统状态空间矩阵,且C pi=C p=[0 1]、D pi=D p=0;R n、R t、R m分别表示维数为n、t、m的实数集,T表示对矩阵进行转置。
所述步骤2包括以下步骤:
步骤2.1:设航空发动机高压涡轮相对换算转速n hi为调度参数θ(i),i=1,2,3,…,N;
步骤2.2:将航空发动机连续小偏差状态变量模型的系统矩阵A p(θ)和控制矩阵B p(θ)表述成仿射参数依赖形式,如下:
A p(θ)=A 0+θA 1,B p(θ)=B 0+θB 1        (2)
其中,A 0、A 1、B 0、B 1分别表示待求的系数矩阵。
将式(2)改写为
Figure PCTCN2018120243-appb-000003
其中,I是单位矩阵。
则有
Figure PCTCN2018120243-appb-000004
其中,[I θI] +为[I θI]的Moore-Penrose伪逆,即求得仿射参数依赖的航空发动机LPV模型的系统矩阵A p(θ)和控制矩阵B p(θ);
步骤2.3:建立仿射参数依赖的航空发动机LPV模型
Figure PCTCN2018120243-appb-000005
所述步骤3建立航空发动机LPV故障估计器的H 综合框架包括以下步骤:
步骤3.1:将存在扰动及传感器与执行机构故障的仿射参数依赖的航空发动机LPV模型P(s,θ)表示为
Figure PCTCN2018120243-appb-000006
其中,d∈R q为扰动信号,f∈R l为故障信号,包括传感器故障及执行机构故障,R q、R l分别表示维数为q、l的实数集;E p、F p、G p、H p是系统状态空间矩阵,P(s,θ)的上LFT结构表示为
Figure PCTCN2018120243-appb-000007
其中,外部输入变量w=[u T d T f T] T∈R p1,w θ∈R r为时变部分Δ(θ)=θI的输出变量,z θ∈R r为时变部分Δ(θ)=θI的输入变量,A p、B 、B pw、C 、C pw、D pθθ、D pθw、D pwθ、D pww是系统状态空间矩阵;R p1、R r分别表示维数为p1、r的实数集,且有p1=t+q+l,即外部输入变量w的维数p1等于航空发动机的输入变量u的维数t、扰动信号d的维数q和故障信号f的维数l之和。
步骤3.2:设故障估计器K(s,θ)形式如下
Figure PCTCN2018120243-appb-000008
其中,x K∈R k为故障估计器K(s,θ)的状态变量,
Figure PCTCN2018120243-appb-000009
表示x K的一阶导数,R k表示维数为k的实数集;u K=[u T y p T] T∈R p2为K(s,θ)的输入变量,p2=t+m,即K(s,θ)的输入变量u K的维数p2等于航空发动机的输入变量u的维数t和航空发动机的输出变量y p的维数m之和;
Figure PCTCN2018120243-appb-000010
的输出变量,即故障信号f的估计值,A K(θ)、B K(θ)、C K(θ)、D K(θ)是系统状态空间矩阵,将K(s,θ)表示成下LFT结构,如下:
Figure PCTCN2018120243-appb-000011
其中,w K∈R r为时变部分Δ K(θ)=θI的输出变量,z K∈R r为时变部分Δ K(θ)=θI的输入 变量,A K、B K1、B 、C K1、C 、D K11、D K1θ、D Kθ1、D Kθθ是系统状态空间矩阵;
步骤3.3:根据航空发动机LPV模型P(s,θ)中时变部分Δ(θ)和故障估计器K(s,θ)中时变部分Δ K(θ),LPV故障估计器的H 综合框架表示为
Figure PCTCN2018120243-appb-000012
其中,
Figure PCTCN2018120243-appb-000013
为故障估计误差,系统矩阵
Figure PCTCN2018120243-appb-000014
系统矩阵
Figure PCTCN2018120243-appb-000015
系统矩阵
Figure PCTCN2018120243-appb-000016
系统矩阵
Figure PCTCN2018120243-appb-000017
系统矩阵
Figure PCTCN2018120243-appb-000018
系统矩阵
Figure PCTCN2018120243-appb-000019
系统矩阵
Figure PCTCN2018120243-appb-000020
系统矩阵
Figure PCTCN2018120243-appb-000021
系统矩阵
Figure PCTCN2018120243-appb-000022
故障估计器矩阵
Figure PCTCN2018120243-appb-000023
矩阵
Figure PCTCN2018120243-appb-000024
矩阵
Figure PCTCN2018120243-appb-000025
矩阵
Figure PCTCN2018120243-appb-000026
矩阵
Figure PCTCN2018120243-appb-000027
矩阵
Figure PCTCN2018120243-appb-000028
矩阵
Figure PCTCN2018120243-appb-000029
矩阵
Figure PCTCN2018120243-appb-000030
矩阵
Figure PCTCN2018120243-appb-000031
矩阵
Figure PCTCN2018120243-appb-000032
矩阵
Figure PCTCN2018120243-appb-000033
矩阵
Figure PCTCN2018120243-appb-000034
矩阵
Figure PCTCN2018120243-appb-000035
矩阵
Figure PCTCN2018120243-appb-000036
矩阵
Figure PCTCN2018120243-appb-000037
矩阵D 04=D 11,矩阵A=A p,矩阵B θ=B ,矩阵B 1=B pw,矩阵B 2=0 n×l,矩阵C θ=C ,矩阵D θθ=D pθθ,矩阵D θ1=D pθw,矩阵D θ2=0 r×l,矩阵C 1=0 p1×n,矩阵D =0 p1×r,矩阵
Figure PCTCN2018120243-appb-000038
矩阵
Figure PCTCN2018120243-appb-000039
矩阵
Figure PCTCN2018120243-appb-000040
矩阵
Figure PCTCN2018120243-appb-000041
矩阵
Figure PCTCN2018120243-appb-000042
矩阵D 22=0 p2×l;n表示航空发动机的状态变量x p的维数,r表示时变部分Δ(θ)的输出变量w θ和时变部分Δ K(θ)的输出变量w K的维数,k表示故障估计器K(s,θ)的状态变量x K的维数。
所述步骤4得到故障估计器存在的有解条件包括以下步骤:
步骤4.1:得到故障估计器K(s,θ)存在的有解条件,即
Figure PCTCN2018120243-appb-000043
Figure PCTCN2018120243-appb-000044
其中,X为正定对称矩阵,全块标量矩阵
Figure PCTCN2018120243-appb-000045
为对称矩阵,γ>0为性能指标;Q、S、R分别表示P的子标量矩阵块。
步骤4.2:对正定对称矩阵X及其逆矩阵X -1进行分块
Figure PCTCN2018120243-appb-000046
其中,L、M、E分别表示X的矩阵块,J、N、F分别表示X -1的子矩阵块。
对全块标量矩阵P及其逆矩阵
Figure PCTCN2018120243-appb-000047
进行分块
Figure PCTCN2018120243-appb-000048
其中,Q 1、Q 2、Q 3分别表示Q的子矩阵块,S 1、S 2、S 3、S 4分别表示S的子矩阵块,R 1、R 2、R 3分别表示R的子矩阵块,
Figure PCTCN2018120243-appb-000049
分别表示
Figure PCTCN2018120243-appb-000050
的子矩阵块,
Figure PCTCN2018120243-appb-000051
分别表示
Figure PCTCN2018120243-appb-000052
的子矩阵块,
Figure PCTCN2018120243-appb-000053
分别表示
Figure PCTCN2018120243-appb-000054
的子矩阵块,
Figure PCTCN2018120243-appb-000055
分别表示
Figure PCTCN2018120243-appb-000056
的子矩阵块。
化简故障估计器K(s,θ)存在的有解条件,即
Figure PCTCN2018120243-appb-000057
Figure PCTCN2018120243-appb-000058
Figure PCTCN2018120243-appb-000059
R>0,Q=-R,S+S T=0        (18)
其中,N L和N J分别代表[C 2 D  D 21]和
Figure PCTCN2018120243-appb-000060
的核空间的基;
步骤4.3:求解线性矩阵不等式(15)-(18),得到矩阵解L、J、Q 3
Figure PCTCN2018120243-appb-000061
S 4
Figure PCTCN2018120243-appb-000062
所述步骤5结合LFT结构设计故障估计器包括以下步骤:
步骤5.1:根据求得的矩阵解L、J、Q 3
Figure PCTCN2018120243-appb-000063
S 4
Figure PCTCN2018120243-appb-000064
由式(13)、(14)求得正定对称矩阵X、全块标量矩阵P及其逆矩阵
Figure PCTCN2018120243-appb-000065
步骤5.2:根据Schur补定理,将线性矩阵不等式(11)表示为
Figure PCTCN2018120243-appb-000066
求解线性矩阵不等式(19),得到故障估计器矩阵Ω;
步骤5.3:得到故障估计器K(s,θ)的状态空间矩阵
Figure PCTCN2018120243-appb-000067
本发明的有益效果为:通过本发明设计的航空发动机传感器及执行机构故障诊断方法,将航空发动机LPV模型和故障估计器分别转化为时不变部分和时变部分组成的LFT结构,其中时变部分随着时变参数向量的变化而变化,因而使故障估计器具有增益调度特性,可以实现外界扰动和建模误差等不确定性条件影响下故障信号的准确估计,从而方便了解故障的类型、产生时间以及严重程度等信息。此外,本发明通过S过程,降低了故障估计器设计的保守性。
附图说明
图1为H=0,Ma=0,n 2=90%工作状态下航空发动机状态空间模型的高压涡轮相对换算转速响应Y nh与试验数据对比曲线。
图2为H=0,Ma=0,n 2=90%工作状态下航空发动机LPV模型的高压涡轮相对换算转速响应Y nh与试验数据对比曲线。
图3为航空发动机LPV模型P(s,θ)的上LFT结构图。
图4为LFT框架下的系统结构图。
图5为LPV故障估计器的H 综合框架。
图6(a)和图6(b)为突变故障估计仿真结果。
图7(a)和图7(b)为缓变故障估计仿真结果。
图8(a)和图8(b)为间歇故障估计仿真结果。
图9为本发明的流程示意图。
具体实施方式
下面结合附图及技术方案对本发明实施例做进一步详细说明。
本发明的流程示意图如图9所示,具体步骤如下:
步骤1.1:向航空发动机输入稳态工作点下的油压p f,待航空发动机高压涡轮相对换算转速n h达到相应稳定状态后,将幅值为0.01p f的油压阶跃信号U pf1输入到航空发动机中,分别采集航空发动机输出高压涡轮相对换算转速响应Y nh1和低压涡轮相对换算转速响应Y nl1
步骤1.2:重复上述过程13次,即在(H=0,Ma=0,n h=88%、89%、…、100%)13个工况的平衡点处分别采集得到给定油压p fi下的高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli,i=1,2,3,…,13。
步骤1.3:以油压阶跃信号U pfi为输入变量,高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli为状态变量,则航空发动机离散型小偏差状态变量模型表示为
Figure PCTCN2018120243-appb-000068
其中,状态变量x p=[Y nl Y nh] T∈R n,输入变量u=U pf∈R t,输出变量y p=Y nh∈R m,i=1,2,3,…,13,下标k、k+1为对应的采样时刻,E i、F i、G i、H i是适当维数的系统状态空间矩阵;R n、R t、R m分别表示维数为n、t、m的实数集,T表示对矩阵进行转置。根据拟合法基本思想,针对式(21)建立线性最小二乘问题,利用MATLAB中的lsqnonlin函数求解其系统矩阵E i,F i,G i,H i
步骤1.4:根据采样周期T=25ms,将各稳态工作点下的航空发动机离散小偏差状态变量模型转换为连续小偏差状态变量模型,得到航空发动机状态变量模型;
Figure PCTCN2018120243-appb-000069
其中A pi、B pi、C pi、D pi是适当维数的系统状态空间矩阵,且C pi=C p=[0 1]、D pi=D p=0,给出在工作点H=0,Ma=0,n 2=90%处的状态空间模型的高压涡轮相对换算转速响应Y nh曲线,如图1所示,其与试验数据的平均相对误差为0.26%。
步骤2.1:设航空发动机高压涡轮相对换算转速n hi为调度参数θ(i),i=1,2,3,…,13。
步骤2.2:将航空发动机连续小偏差状态变量模型系统矩阵A p(θ)和控制矩阵B p(θ)表述成仿射参数依赖形式,如下:
Figure PCTCN2018120243-appb-000070
其中A 0、A 1、B 0、B 1分别表示待求的系数矩阵。
将式(23)改写为
Figure PCTCN2018120243-appb-000071
其中,I是单位矩阵。
则有
Figure PCTCN2018120243-appb-000072
利用MATLAB中的pinv函数求得
Figure PCTCN2018120243-appb-000073
的Moore-Penrose伪逆
Figure PCTCN2018120243-appb-000074
并对 变参数θ进行变量变换,使其θ∈[-1,1],则可求得
Figure PCTCN2018120243-appb-000075
步骤2.3:建立仿射参数依赖的航空发动机LPV模型
Figure PCTCN2018120243-appb-000076
给出在工作点H=0,Ma=0,n 2=90%处航空发动机LPV模型的高压涡轮相对换算转速响应Y nh曲线,如图2所示,其与试验数据的平均相对误差为2.51%。
步骤3.1:将存在扰动及传感器与执行机构故障的仿射参数依赖的航空发动机LPV模型P(s,θ)表示为
Figure PCTCN2018120243-appb-000077
其中,d∈R q为扰动信号,取标准差0.001的高斯白噪声,f∈R l为故障信号,包括传感器故障及执行器故障,分别取突变故障、缓变故障和间歇故障,R q、R l分别表示维数为q、l的实数集;
Figure PCTCN2018120243-appb-000078
G p=0.2,H p=1。
P(s,θ)的上LFT结构可表示为下式,具体的如图3所示,
Figure PCTCN2018120243-appb-000079
其中,F u表示上LFT结构,P′表示P(s,θ)中时不变部分,Δ(θ)=θI表示P(s,θ)中时变部分,即
Figure PCTCN2018120243-appb-000080
其中,外部输入变量w=[u T d T f T] T∈R p1,w θ∈R r为时变部分Δ(θ)=θI的输出变量,z θ∈R r为时变部分Δ(θ)=θI的输入变量;R p1、R r分别表示维数为p1、r的实数集,且有p1=t+q+l,即外部输入变量w的维数p1等于航空发动机的输入变量u的维数t、扰动信号d的维数q和故障信号f的维数l之和;系统状态空间矩阵为
Figure PCTCN2018120243-appb-000081
步骤3.2:设故障估计器K(s,θ)形式如下
Figure PCTCN2018120243-appb-000082
其中,xK∈R k为故障估计器K(s,θ)的状态变量,u K=[u T y p T] T∈R p2为K(s,θ)的输入变量,p2=t+m,即K(s,θ)的输入变量u K的维数p2等于航空发动机的输入变量u的维数t和航空发动机的输出变量y p的维数m之和;
Figure PCTCN2018120243-appb-000083
为K(s,θ)的输出变量,即故障信号f的估计值,A K(θ)、B K(θ)、C K(θ)、D K(θ)是系统状态空间矩阵。将K(s,θ)表示成下LFT结构,如下:
Figure PCTCN2018120243-appb-000084
其中,F l表示下LFT结构,K′表示K(s,θ)中时不变部分,Δ K(θ)=θI表示K(s,θ)中时变部分,即
Figure PCTCN2018120243-appb-000085
其中,w K∈R r为时变部分Δ K(θ)=θI的输出变量,z K∈R r为时变部分Δ K(θ)=θI的输入变量,A K、B K1、B 、C K1、C 、D K11、D K1θ、D Kθ1、D Kθθ是适当维数的系统状态空间矩阵。
步骤3.3:LFT框架下的系统连接图如图4所示,则图4中系统P 1的状态空间表达式为
Figure PCTCN2018120243-appb-000086
其中,系统矩阵A=A p,系统矩阵B θ=B ,系统矩阵B 1=B pw,系统矩阵B 2=0 n×l,系统矩阵C θ=C ,系统矩阵D θθ=D pθθ,系统矩阵D θ1=D pθw,系统矩阵D θ2=0 r×l,系统矩阵C 1=0 p1×n,系统矩阵D =0 p1×r,系统矩阵
Figure PCTCN2018120243-appb-000087
系统矩阵
Figure PCTCN2018120243-appb-000088
系统矩阵
Figure PCTCN2018120243-appb-000089
系统矩阵
Figure PCTCN2018120243-appb-000090
系统矩阵
Figure PCTCN2018120243-appb-000091
D 22=0 p2×l;n表示航空发动机的状态变量x p的维数,r表示时变部分Δ(θ)的输出变量w θ和时变部分Δ K(θ)的输出变量w K的维数。
根据航空发动机LPV模型P(s,θ)中时变部分Δ(θ)和故障估计器K(s,θ)中时变部分Δ K(θ),LPV故障估计器的H 综合框架可表示为下式,具体的如图5所示
Figure PCTCN2018120243-appb-000092
其中,
Figure PCTCN2018120243-appb-000093
为故障估计误差,即LPV故障估计器的H 综合框架的输出变量,系统矩阵
Figure PCTCN2018120243-appb-000094
系统矩阵
Figure PCTCN2018120243-appb-000095
系统矩阵
Figure PCTCN2018120243-appb-000096
系统矩阵
Figure PCTCN2018120243-appb-000097
系统矩阵
Figure PCTCN2018120243-appb-000098
系统矩阵
Figure PCTCN2018120243-appb-000099
系统矩阵
Figure PCTCN2018120243-appb-000100
系统矩阵
Figure PCTCN2018120243-appb-000101
系统矩阵
Figure PCTCN2018120243-appb-000102
故障估计器矩阵
Figure PCTCN2018120243-appb-000103
矩阵
Figure PCTCN2018120243-appb-000104
矩阵
Figure PCTCN2018120243-appb-000105
矩阵
Figure PCTCN2018120243-appb-000106
矩阵
Figure PCTCN2018120243-appb-000107
矩阵
Figure PCTCN2018120243-appb-000108
矩阵
Figure PCTCN2018120243-appb-000109
矩阵
Figure PCTCN2018120243-appb-000110
矩阵
Figure PCTCN2018120243-appb-000111
矩阵
Figure PCTCN2018120243-appb-000112
矩阵
Figure PCTCN2018120243-appb-000113
矩阵
Figure PCTCN2018120243-appb-000114
矩阵
Figure PCTCN2018120243-appb-000115
矩阵
Figure PCTCN2018120243-appb-000116
矩阵
Figure PCTCN2018120243-appb-000117
矩阵D 04=D 11
步骤4.1:如果存在一个对称正定矩阵X,对称矩阵
Figure PCTCN2018120243-appb-000118
使得式(37)和式(38)成立,
Figure PCTCN2018120243-appb-000119
Figure PCTCN2018120243-appb-000120
则闭环系统(36)渐进稳定,并且从外部输入w到故障估计误差e f的闭环传递函数的L 2诱导范数小于性能指标γ(γ>0)。即故障估计器K(s,θ)存在的有解条件为式(37)、式(38)。其中,Q、S、R分别表示P的子矩阵块。
步骤4.2:对正定对称矩阵X及其逆矩阵X -1进行分块
Figure PCTCN2018120243-appb-000121
其中,L、M、E分别表示X的矩阵块,J、N、F分别表示X -1的子矩阵块。
由于X为正定对称矩阵,可得
Figure PCTCN2018120243-appb-000122
对全块标量矩阵P及其逆矩阵
Figure PCTCN2018120243-appb-000123
进行分块
Figure PCTCN2018120243-appb-000124
其中,Q 1、Q 2、Q 3分别表示Q的子矩阵块,S 1、S 2、S 3、S 4分别表示S的子矩阵块,R 1、R 2、R 3分别表示R的子矩阵块,
Figure PCTCN2018120243-appb-000125
分别表示
Figure PCTCN2018120243-appb-000126
的子矩阵块,
Figure PCTCN2018120243-appb-000127
分别表示
Figure PCTCN2018120243-appb-000128
的子矩阵块,
Figure PCTCN2018120243-appb-000129
分别表示
Figure PCTCN2018120243-appb-000130
的子矩阵块,
Figure PCTCN2018120243-appb-000131
分别表示
Figure PCTCN2018120243-appb-000132
的子矩阵块。
线性矩阵不等式(37)整理为
Figure PCTCN2018120243-appb-000133
其中,矩阵
Figure PCTCN2018120243-appb-000134
矩阵
Figure PCTCN2018120243-appb-000135
矩阵V=[T 2 T 3 T 4],矩阵Γ=Ω。
若要满足式(38)成立,需要验证在变参数θ所有可能的轨迹上都成立,这是不可能实现的,为此,限定全块标量矩阵
Figure PCTCN2018120243-appb-000136
的结构使其自成立。对于每一个变参数θ,当R≥0时有下式成立
Figure PCTCN2018120243-appb-000137
因此,令Q=-R,S+S T=0。即式(38)可整理为
R>0,Q=-R,S+S T=0          (44)
综上所述,故障估计器K(s,θ)存在的有解条件转化为式(40)、式(42)、式(44)。
步骤4.3:线性矩阵不等式(42)整理为
Figure PCTCN2018120243-appb-000138
Figure PCTCN2018120243-appb-000139
其中,U 和V 分别是U T和V的核空间的基。
Figure PCTCN2018120243-appb-000140
经过简单计算,线性矩阵不等式(45)、(46)化简为
Figure PCTCN2018120243-appb-000141
Figure PCTCN2018120243-appb-000142
其中,N L和N J分别代表[C 2 D  D 21]和
Figure PCTCN2018120243-appb-000143
的核空间的基。
步骤4.4:利用MATLAB里的LMI工具箱求解线性矩阵不等式(40)、(44)、(47)、(48),得到最优的γ取值为0.21,以及相应的矩阵解L、J、Q 3
Figure PCTCN2018120243-appb-000144
S 4
Figure PCTCN2018120243-appb-000145
步骤5.1:根据求得的矩阵解L、J、Q 3
Figure PCTCN2018120243-appb-000146
S 4
Figure PCTCN2018120243-appb-000147
由式(39)、(41)求得正定对称矩阵X、全块标量矩阵P及其逆矩阵
Figure PCTCN2018120243-appb-000148
步骤5.2:根据Schur补定理,线性矩阵不等式(37)表示为
Figure PCTCN2018120243-appb-000149
将闭环系统(36)中的值代入,可得
Figure PCTCN2018120243-appb-000150
其中,
Figure PCTCN2018120243-appb-000151
Figure PCTCN2018120243-appb-000152
求解线性矩阵不等式(50),得到故障估计器矩阵Ω。
步骤5.3:得到故障估计器K(s,θ)的状态空间矩阵
Figure PCTCN2018120243-appb-000153
在工作点H=0km,Ma=0,n 2=90%处的仿真结果如图6(a)和图6(b)、图7(a)和图7(b)、图8(a)和图8(b)所示,并与标准H 方法进行比较。仿真结果表明标准H 方法设计的固定参数故障估计器不能很好地应对变参数的变化,而本发明设计的LPV故障估计器可以迅速地检测系统中的故障,并准确地重构故障信号,具有明显的性能优势。

Claims (1)

  1. 一种基于LFT的航空发动机传感器及执行机构故障诊断方法,其特征在于,包括以下步骤:
    步骤1:采用小扰动法与线性拟合法相结合的方法建立航空发动机状态变量模型;
    步骤1.1:向航空发动机输入稳态工作点下的油压p f,待航空发动机高压涡轮相对换算转速n h达到相应稳定状态后,将幅值为0.01p f的油压阶跃信号U pf1输入到航空发动机中,分别采集航空发动机输出高压涡轮相对换算转速响应Y nh1和低压涡轮相对换算转速响应Y nl1
    步骤1.2:重复步骤1的过程N次,分别采集得到给定油压p fi下的高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli,i=1,2,3,…,N;
    步骤1.3:以油压阶跃信号U pfi为输入变量,高压涡轮相对换算转速响应Y nhi和低压涡轮相对换算转速响应Y nli为状态变量,根据线性拟合法求解各稳态工作点下的航空发动机离散小偏差状态变量模型;
    步骤1.4:根据采样周期T,将各稳态工作点下的航空发动机离散小偏差状态变量模型转换为连续小偏差状态变量模型,得到航空发动机状态变量模型;
    Figure PCTCN2018120243-appb-100001
    其中,状态变量x p=[Y nl T Y nh T] T∈R n
    Figure PCTCN2018120243-appb-100002
    表示x p的一阶导数,输入变量u=U pf∈R t,输出变量y p=Y nh∈R m,A pi、B pi、C pi、D pi是系统状态空间矩阵,且C pi=C p=[0 1]、D pi=D p=0;R n、R t、R m分别表示维数为n、t、m的实数集,T表示对矩阵进行转置;
    步骤2:建立仿射参数依赖的航空发动机LPV模型;
    步骤2.1:设航空发动机高压涡轮相对换算转速n hi为调度参数θ(i),i=1,2,3,…,N;
    步骤2.2:将航空发动机连续小偏差状态变量模型的系统矩阵A p(θ)和控制矩阵B p(θ)表述成仿射参数依赖形式,如下:
    A p(θ)=A 0+θA 1,B p(θ)=B 0+θB 1    (2)
    其中,A 0、A 1、B 0、B 1分别表示待求的系数矩阵;
    将式(2)改写为
    Figure PCTCN2018120243-appb-100003
    其中,I是单位矩阵;
    则有
    Figure PCTCN2018120243-appb-100004
    其中,[I θI] +为[I θI]的Moore-Penrose伪逆,即求得仿射参数依赖的航空发动机LPV模型的系统矩阵A p(θ)和控制矩阵B p(θ);
    步骤2.3:建立仿射参数依赖的航空发动机LPV模型
    Figure PCTCN2018120243-appb-100005
    步骤3:将存在扰动及传感器与执行机构故障的仿射参数依赖的航空发动机LPV模型转化为LFT结构,建立航空发动机LPV故障估计器的H 综合框架;
    步骤3.1:将存在扰动及传感器与执行机构故障的仿射参数依赖的航空发动机LPV模型P(s,θ)表示为
    Figure PCTCN2018120243-appb-100006
    其中,d∈R q为扰动信号,f∈R l为故障信号,包括传感器故障及执行机构故障,R q、R l分别表示维数为q、l的实数集;E p、F p、G p、H p是系统状态空间矩阵,P(s,θ)的上LFT结构表示为
    Figure PCTCN2018120243-appb-100007
    其中,外部输入变量w=[u T d T f T] T∈R p1,w θ∈R r为时变部分Δ(θ)=θI的输出变量,z θ∈R r为时变部分Δ(θ)=θI的输入变量,A p、B 、B pw、C 、C pw、D pθθ、D pθw、D pwθ、D pww是系统状态空间矩阵;R p1、R r分别表示维数为p1、r的实数集,且有p1=t+q+l,即外部输入变量w的维数p1等于航空发动机的输入变量u的维数t、扰动信号d的维数q和故障信号f的维数l之和;
    步骤3.2:设故障估计器K(s,θ)形式如下
    Figure PCTCN2018120243-appb-100008
    其中,x K∈R k为故障估计器K(s,θ)的状态变量,
    Figure PCTCN2018120243-appb-100009
    表示x K的一阶导数,R k表示维数为k的实数集;u K=[u T y p T] T∈R p2为K(s,θ)的输入变量,p2=t+m,即K(s,θ)的输入变量u K的维数p2等于航空发动机的输入变量u的维数t和航空发动机的输出变量y p的维数m之和;
    Figure PCTCN2018120243-appb-100010
    为K(s,θ)的输出变量,即故障信号f的估计值,A K(θ)、B K(θ)、C K(θ)、D K(θ)是系统状态空间矩阵,将K(s,θ)表示成下LFT结构,如下:
    Figure PCTCN2018120243-appb-100011
    其中,w K∈R r为时变部分Δ K(θ)=θI的输出变量,z K∈R r为时变部分Δ K(θ)=θI的输入变量,A K、B K1、B 、C K1、C 、D K11、D K1θ、D Kθ1、D Kθθ是系统状态空间矩阵;
    步骤3.3:根据航空发动机LPV模型P(s,θ)中时变部分Δ(θ)和故障估计器K(s,θ)中时变部分Δ K(θ),LPV故障估计器的H 综合框架表示为
    Figure PCTCN2018120243-appb-100012
    其中,
    Figure PCTCN2018120243-appb-100013
    为故障估计误差,系统矩阵
    Figure PCTCN2018120243-appb-100014
    系统矩阵
    Figure PCTCN2018120243-appb-100015
    系统矩阵
    Figure PCTCN2018120243-appb-100016
    系统矩阵
    Figure PCTCN2018120243-appb-100017
    系统矩阵
    Figure PCTCN2018120243-appb-100018
    系统矩阵
    Figure PCTCN2018120243-appb-100019
    系统矩阵
    Figure PCTCN2018120243-appb-100020
    系统矩阵
    Figure PCTCN2018120243-appb-100021
    系统矩阵
    Figure PCTCN2018120243-appb-100022
    故障估计器矩阵
    Figure PCTCN2018120243-appb-100023
    矩阵
    Figure PCTCN2018120243-appb-100024
    矩阵
    Figure PCTCN2018120243-appb-100025
    矩阵
    Figure PCTCN2018120243-appb-100026
    矩阵
    Figure PCTCN2018120243-appb-100027
    矩阵
    Figure PCTCN2018120243-appb-100028
    矩阵
    Figure PCTCN2018120243-appb-100029
    矩阵
    Figure PCTCN2018120243-appb-100030
    矩阵
    Figure PCTCN2018120243-appb-100031
    矩阵
    Figure PCTCN2018120243-appb-100032
    矩阵
    Figure PCTCN2018120243-appb-100033
    矩阵
    Figure PCTCN2018120243-appb-100034
    矩阵
    Figure PCTCN2018120243-appb-100035
    矩阵
    Figure PCTCN2018120243-appb-100036
    矩阵
    Figure PCTCN2018120243-appb-100037
    矩阵D 04=D 11,矩阵A=A p,矩阵B θ=B ,矩阵B 1=B pw,矩阵B 2=0 n×l,矩阵C θ=C ,矩阵D θθ=D pθθ,矩阵D θ1=D pθw,矩阵D θ2=0 r×l,矩阵C 1=0 p1×n,矩阵D =0 p1×r,矩阵
    Figure PCTCN2018120243-appb-100038
    矩阵
    Figure PCTCN2018120243-appb-100039
    矩阵
    Figure PCTCN2018120243-appb-100040
    矩阵
    Figure PCTCN2018120243-appb-100041
    矩阵
    Figure PCTCN2018120243-appb-100042
    D 22=0 p2×l;n表示航空发动机的状态变量x p的维数,r表示时变部分Δ(θ)的输出变量w θ和时变部分Δ K(θ)的输出变量w K的维数,k表示故障估计器K(s,θ)的状态变量x K的维数;
    步骤4:求解一组线性矩阵不等式LMIs,获得故障估计器存在的有解条件;
    步骤4.1:得到故障估计器K(s,θ)存在的有解条件,即
    Figure PCTCN2018120243-appb-100043
    Figure PCTCN2018120243-appb-100044
    其中,X为正定对称矩阵,全块标量矩阵
    Figure PCTCN2018120243-appb-100045
    为对称矩阵,γ>0为性能指标;Q、S、R分别表示P的子矩阵块;
    步骤4.2:对正定对称矩阵X及其逆矩阵X -1进行分块
    Figure PCTCN2018120243-appb-100046
    其中,L、M、E分别表示X的矩阵块,J、N、F分别表示X -1的子矩阵块;
    对全块标量矩阵P及其逆矩阵
    Figure PCTCN2018120243-appb-100047
    进行分块
    Figure PCTCN2018120243-appb-100048
    其中,Q 1、Q 2、Q 3分别表示Q的子矩阵块,S 1、S 2、S 3、S 4分别表示S的子矩阵块,R 1、R 2、R 3分别表示R的子矩阵块,
    Figure PCTCN2018120243-appb-100049
    分别表示
    Figure PCTCN2018120243-appb-100050
    的子矩阵块,
    Figure PCTCN2018120243-appb-100051
    分别表示
    Figure PCTCN2018120243-appb-100052
    的子矩阵块,
    Figure PCTCN2018120243-appb-100053
    分别表示
    Figure PCTCN2018120243-appb-100054
    的子矩阵块,
    Figure PCTCN2018120243-appb-100055
    分别表示
    Figure PCTCN2018120243-appb-100056
    的子矩阵块;
    化简故障估计器K(s,θ)存在的有解条件,即
    Figure PCTCN2018120243-appb-100057
    Figure PCTCN2018120243-appb-100058
    Figure PCTCN2018120243-appb-100059
    R>0,Q=-R,S+S T=0    (18)
    其中,N L和N J分别代表[C 2 D  D 21]和
    Figure PCTCN2018120243-appb-100060
    的核空间的基;
    步骤4.3:求解线性矩阵不等式(15)-(18),得到矩阵解L、J、Q 3
    Figure PCTCN2018120243-appb-100061
    S 4
    Figure PCTCN2018120243-appb-100062
    步骤5:结合LFT结构设计故障估计器,实现航空发动机传感器及执行机构的故障诊断;
    步骤5.1:根据求得的矩阵解L、J、Q 3
    Figure PCTCN2018120243-appb-100063
    S 4
    Figure PCTCN2018120243-appb-100064
    由式(13)、(14)求得正定对称矩阵X、全块标量矩阵P及其逆矩阵
    Figure PCTCN2018120243-appb-100065
    步骤5.2:根据Schur补定理,将线性矩阵不等式(11)表示为
    Figure PCTCN2018120243-appb-100066
    求解线性矩阵不等式(19),得到故障估计器矩阵Ω;
    步骤5.3:得到故障估计器K(s,θ)的状态空间矩阵
    Figure PCTCN2018120243-appb-100067
PCT/CN2018/120243 2018-12-11 2018-12-11 一种基于lft的航空发动机传感器及执行机构故障诊断方法 WO2020118512A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US16/604,042 US11203446B2 (en) 2018-12-11 2018-12-11 Method for fault diagnosis of aero-engine sensor and actuator based on LFT
PCT/CN2018/120243 WO2020118512A1 (zh) 2018-12-11 2018-12-11 一种基于lft的航空发动机传感器及执行机构故障诊断方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/120243 WO2020118512A1 (zh) 2018-12-11 2018-12-11 一种基于lft的航空发动机传感器及执行机构故障诊断方法

Publications (1)

Publication Number Publication Date
WO2020118512A1 true WO2020118512A1 (zh) 2020-06-18

Family

ID=71077065

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/120243 WO2020118512A1 (zh) 2018-12-11 2018-12-11 一种基于lft的航空发动机传感器及执行机构故障诊断方法

Country Status (2)

Country Link
US (1) US11203446B2 (zh)
WO (1) WO2020118512A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113189968A (zh) * 2021-05-08 2021-07-30 哈尔滨工业大学 互联工业过程的分布式故障诊断方法
CN113282004A (zh) * 2021-05-20 2021-08-20 南京航空航天大学 基于神经网络的航空发动机线性变参数模型建立方法
CN113985734A (zh) * 2020-12-08 2022-01-28 北京航空航天大学 基于执行器故障的高超声速导弹的自适应容错控制器
CN114779735A (zh) * 2022-04-02 2022-07-22 西北工业大学 航空发动机分布式故障诊断系统硬件在环仿真平台及方法
CN115437359A (zh) * 2022-11-07 2022-12-06 西北工业大学 一种面向传感器故障的飞机容错控制方法、系统及装置

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3862835B1 (en) * 2020-02-10 2023-10-25 Volocopter GmbH Method and system for monitoring a condition of a vtol-aircraft
CN112733320B (zh) * 2020-12-09 2022-03-22 淮阴工学院 一种基于delta算子的升压变换器执行器故障检测方法
CN112799374B (zh) * 2020-12-24 2023-01-10 南京财经大学 一种Delta算子切换粮食管理系统的全阶故障估计观测器的设计方法
CN113110049B (zh) * 2021-04-13 2022-04-01 龙泉东土青瓷有限公司 一种用于青瓷素烧炉窑温度监测的可靠估计方法
CN114859855B (zh) * 2022-04-22 2023-03-14 大连理工大学 基于参数依赖Lyapunov函数的汽车发动机LPV系统故障诊断装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110054863A1 (en) * 2009-09-03 2011-03-03 Adaptics, Inc. Method and system for empirical modeling of time-varying, parameter-varying, and nonlinear systems via iterative linear subspace computation
CN103983453A (zh) * 2014-05-08 2014-08-13 南京航空航天大学 一种航空发动机的执行机构和传感器故障诊断的区分方法
CN104950876A (zh) * 2015-06-15 2015-09-30 西北工业大学 基于故障敏感性约束的航天器弱小故障检测方法
CN106873568A (zh) * 2015-12-11 2017-06-20 中国航空工业第六八研究所 基于h无穷鲁棒未知输入观测器的传感器故障诊断方法
CN107703742A (zh) * 2017-09-06 2018-02-16 南京邮电大学 一种柔性航天器传感器故障调节方法
CN107797543A (zh) * 2017-09-26 2018-03-13 大连理工大学 一种航空发动机燃油调节器故障诊断方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050193739A1 (en) * 2004-03-02 2005-09-08 General Electric Company Model-based control systems and methods for gas turbine engines
US8668434B2 (en) * 2009-09-02 2014-03-11 United Technologies Corporation Robust flow parameter model for component-level dynamic turbine system control
US9073623B1 (en) * 2013-03-15 2015-07-07 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration System and method for dynamic aeroelastic control

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110054863A1 (en) * 2009-09-03 2011-03-03 Adaptics, Inc. Method and system for empirical modeling of time-varying, parameter-varying, and nonlinear systems via iterative linear subspace computation
CN103983453A (zh) * 2014-05-08 2014-08-13 南京航空航天大学 一种航空发动机的执行机构和传感器故障诊断的区分方法
CN104950876A (zh) * 2015-06-15 2015-09-30 西北工业大学 基于故障敏感性约束的航天器弱小故障检测方法
CN106873568A (zh) * 2015-12-11 2017-06-20 中国航空工业第六八研究所 基于h无穷鲁棒未知输入观测器的传感器故障诊断方法
CN107703742A (zh) * 2017-09-06 2018-02-16 南京邮电大学 一种柔性航天器传感器故障调节方法
CN107797543A (zh) * 2017-09-26 2018-03-13 大连理工大学 一种航空发动机燃油调节器故障诊断方法

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113985734A (zh) * 2020-12-08 2022-01-28 北京航空航天大学 基于执行器故障的高超声速导弹的自适应容错控制器
CN113985734B (zh) * 2020-12-08 2024-04-12 北京航空航天大学 基于执行器故障的高超声速导弹的自适应容错控制器
CN113189968A (zh) * 2021-05-08 2021-07-30 哈尔滨工业大学 互联工业过程的分布式故障诊断方法
CN113282004A (zh) * 2021-05-20 2021-08-20 南京航空航天大学 基于神经网络的航空发动机线性变参数模型建立方法
CN113282004B (zh) * 2021-05-20 2022-06-10 南京航空航天大学 基于神经网络的航空发动机线性变参数模型建立方法
CN114779735A (zh) * 2022-04-02 2022-07-22 西北工业大学 航空发动机分布式故障诊断系统硬件在环仿真平台及方法
CN114779735B (zh) * 2022-04-02 2024-04-26 西北工业大学 航空发动机分布式故障诊断系统硬件在环仿真平台及方法
CN115437359A (zh) * 2022-11-07 2022-12-06 西北工业大学 一种面向传感器故障的飞机容错控制方法、系统及装置
CN115437359B (zh) * 2022-11-07 2023-02-28 西北工业大学 一种面向传感器故障的飞机容错控制方法、系统及装置

Also Published As

Publication number Publication date
US11203446B2 (en) 2021-12-21
US20200339285A1 (en) 2020-10-29

Similar Documents

Publication Publication Date Title
WO2020118512A1 (zh) 一种基于lft的航空发动机传感器及执行机构故障诊断方法
CN109799803B (zh) 一种基于lft的航空发动机传感器及执行机构故障诊断方法
WO2021027093A1 (zh) 一种涡扇发动机控制系统主动容错控制方法
CN107703742B (zh) 一种柔性航天器传感器故障调节方法
Badihi et al. Fault-tolerant individual pitch control for load mitigation in wind turbines with actuator faults
CN106125550B (zh) 一种高铁牵引整流器的复合故障估计与容错控制方法
CN106547207B (zh) 一种非线性多输入多输出系统混合式观测器构建方法
Chen et al. Active FTC for hydraulic pitch system for an off-shore wind turbine
CN103529830B (zh) 一种基于有限频域飞行控制系统渐变故障的诊断设计方法
CN103472820A (zh) 一种基于偏最小二乘算法的推进系统故障诊断方法
CN108427288B (zh) 一类具有时变时延的网络化线性参数变化系统的h∞容错控制方法
CN106873568B (zh) 基于h无穷鲁棒未知输入观测器的传感器故障诊断方法
CN112305918A (zh) 超螺旋观测器下的多智能体系统滑模容错一致性控制算法
CN104390657A (zh) 一种发电机组运行参数测量传感器故障诊断方法及系统
CN107340116B (zh) 一种基于时滞补偿的风洞支杆抑振方法
Guo et al. Observer-based optimal fault detection and diagnosis using conditional probability distributions
CN105242544A (zh) 考虑随机扰动的非线性多无人机系统容错编队控制方法
CN102269125A (zh) 风力发电机额定风速以上鲁棒变桨控制器设计方法
Cao et al. Anti-disturbance fault diagnosis for non-Gaussian stochastic distribution systems with multiple disturbances
CN111880403A (zh) 航空发动机最大推力状态容错二自由度μ控制器
CN111859555A (zh) 输入受限的航空发动机最大推力状态鲁棒容错控制器
Zhuo et al. Real-time fault diagnosis for gas turbine blade based on output-hidden feedback elman neural network
Hu et al. Fuzzy adaptive hybrid compensation for compound faults of hypersonic flight vehicle
Wu et al. Variable forgetting factor identification algorithm for fault diagnosis of wind turbines
CN109323841B (zh) 基于网格的机翼总载荷与分布载荷的协调方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18943060

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18943060

Country of ref document: EP

Kind code of ref document: A1