CN111538316A - Performance-based fault diagnosis method and system for closed-loop control system actuators - Google Patents

Abstract

The invention discloses a performance-based fault diagnosis method and system for an actuating mechanism of a closed-loop control system. The method comprises the following steps: establishing a linear nominal model of the control system under the condition of no fault, and determining an identification model of the actual control system according to input data and output data of the actual control system; determining a first time domain performance residual error according to a difference value between the output of the linear nominal model under the closed-loop feedback and the output of the identification model under the closed-loop feedback; determining a first frequency domain performance residual error and a first stable domain performance residual error by adopting a gap measurement method according to the linear nominal model and the identification model; and carrying out fault detection according to the time domain performance residual error, the frequency domain performance residual error and the stable domain performance residual error between the linear nominal model and the identification model. By adopting the method and the system, the fault diagnosis is carried out by establishing the relation between the fault of the actuating mechanism and the performance change of the control system after the closed loop, so that the reliability and the applicability of the fault diagnosis of the control system are improved.

Description

Performance-based fault diagnosis method and system for closed-loop control system actuators

technical field

The invention relates to the technical field of fault diagnosis, in particular to a performance-based method and system for fault diagnosis of an executive mechanism of a closed-loop control system.

Background technique

The traditional fault diagnosis method is generally based on the hardware redundancy system and adopts the voting strategy for diagnosis. The main problem of this kind of hardware redundancy is the high maintenance cost of the equipment and the need for additional installation space. In order to solve this problem, an analytical redundant diagnostic scheme is proposed. The residual signal is generated by the multi-signal consistency test technology, and the fault of the system is determined by analyzing the residual, that is, the model-based fault diagnosis method. Common model-based fault detection and isolation (FDI) methods mainly include observer method, parity relation method and parameter estimation method.

Most of the existing fault diagnosis technologies are for the open-loop system, and do not consider the influence of the control system closed-loop itself on the fault diagnosis performance. In addition, most diagnostic methods assume that the faults in the system are additive faults, which have certain limitations. However, for the closed-loop feedback control system, because the closed-loop control system itself has a certain robustness and fault tolerance, that is, it is not sensitive to system changes (additive or multiplicative changes), and has a wider stable working area. Therefore, using open-loop information for fault diagnosis in the existing diagnostic methods cannot measure the impact of the fault on the performance of the closed-loop feedback control system. If the impact of the fault on the performance of the closed-loop system cannot be known, it will adversely affect the fault decision-making and fault-tolerant control of the system. For example, in some feedback control systems, although the actuator has a very small fault, it will bring great harm from the perspective of the system closed-loop performance. Similarly, in some cases, despite the seemingly serious failure of the actuator (for example, the failure of the actuator makes the open-loop system change from a stable system to an unstable system), the feedback controller itself Under the action of correction, the influence of these faults will be weakened and suppressed, and even the closed-loop performance of the system will not be significantly degraded. Fault-tolerant control measures such as configuration, so as to avoid the risks brought by some unnecessary fault-tolerant switching control. Therefore, it is necessary to carry out fault diagnosis from the indicators at the functional level of the closed-loop control system.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a performance-based method and system for diagnosing faults of an actuator of a closed-loop control system. By establishing the connection between the fault of the actuator and the performance changes of the control system after the closed-loop, the fault diagnosis is performed, and the reliability of the fault diagnosis of the control system is improved. sex and practicality.

For achieving the above object, the present invention provides the following scheme:

A performance-based fault diagnosis method for an actuator of a closed-loop control system, comprising:

Obtain the input data and output data of the actual control system;

According to the working principle of the control system, establish the linear nominal model of the control system model under no fault condition;

Determine the identification model of the actual control system according to the input data and output data of the actual control system;

Get command input data;

Calculate the output of the linear nominal model under closed-loop feedback according to the instruction input data, and calculate the output of the identification model under closed-loop feedback according to the instruction input data;

A first time-domain performance residual is determined according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the identification model under closed-loop feedback; the first time-domain performance residual is a linear nominal model and the time-domain performance residuals between the identification models;

According to the linear nominal model and the identification model, the gap measurement method is used to determine the first frequency domain performance residual and the first stability domain performance residual; the first frequency domain performance residual is the linear nominal model and the identification model. The frequency domain performance residual between the two, the first stability domain performance residual is the stability domain performance residual between the linear nominal model and the identification model;

The fault detection is performed according to the first time domain performance residual, the first frequency domain performance residual and the first stability domain performance residual.

Optionally, the performing fault detection according to the first time-domain performance residual, the first frequency-domain performance residual, and the first stability-domain performance residual specifically includes:

respectively normalizing the first time-domain performance residual, the first frequency-domain performance residual, and the first stability-domain performance residual;

The normalized first time-domain performance residual, the normalized first frequency-domain performance residual, and the normalized first stable-domain performance residual are formed into points in the three-dimensional composite performance residual space; The coordinate system of the three-dimensional composite performance residual space is that the time domain performance residual is the x-axis, the frequency domain performance residual is the y-axis, and the stability domain performance residual is the z-axis, and the points in the coordinate system form the origin. composite performance residual vector;

Determine the size of the composite performance residual vector length and the preset fault threshold; if the composite performance residual vector length is less than the preset fault threshold, the actual control system actuator does not fail; otherwise, the actual control system actuator fails.

Optionally, after performing the fault detection according to the first time-domain performance residual, the first frequency-domain performance residual, and the first stability-domain performance residual, the method further includes:

Obtaining characterization parameters of different failure modes; the characterization parameters are parameters representing failures;

According to the characterization parameters of different failure modes, the linear failure models of the control system model under failure conditions are established respectively;

Calculate the output of the linear fault model under closed-loop feedback according to the instruction input data;

A second time-domain performance residual is determined according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the linear fault model under closed-loop feedback; the second time-domain performance residual is a linear nominal Time-domain performance residuals between the model and the linear fault model;

According to the linear nominal model and the linear fault model, the gap measurement method is used to determine the second frequency domain performance residual and the second stability domain performance residual; the second frequency domain performance residual is the linear nominal model and the linear Frequency domain performance residuals between fault models, the second stability domain performance residuals are stability domain performance residuals between the linear nominal model and the linear fault model;

The fault type and fault degree are judged according to the second time domain performance residual, the second frequency domain performance residual and the second stability domain performance residual.

Optionally, the judgment of the fault type and fault degree according to the second time-domain performance residual, the second frequency-domain performance residual, and the second stability-domain performance residual specifically includes:

For each fault type, the second time domain performance residual, the second frequency domain performance residual and the second stability domain performance residual are respectively normalized;

According to the normalized second time domain performance residual, the normalized second frequency domain performance residual and the normalized second stable domain performance residual of each fault type, the actuator fault space library is established ;

The fault type and fault size are judged in the actuator fault space library according to the direction of the composite performance residual vector.

The present invention also provides a performance-based closed-loop control system actuator fault diagnosis system, comprising:

The actual control system data acquisition module is used to obtain the input data and output data of the actual control system;

The linear nominal model establishment module is used to establish the linear nominal model of the control system model under no fault condition according to the working principle of the control system;

an identification model establishment module, used for determining the identification model of the actual control system according to the input data and output data of the actual control system;

The command input data acquisition module is used to obtain the command input data;

a first output calculation module, configured to calculate the output of the linear nominal model under closed-loop feedback according to the instruction input data, and calculate the output of the identification model under closed-loop feedback according to the instruction input data;

a first time-domain performance residual determination module, configured to determine a first time-domain performance residual according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the identification model under closed-loop feedback; the The first time-domain performance residual is the time-domain performance residual between the linear nominal model and the identification model;

a first gap metric calculation module, configured to use a gap metric method to determine a first frequency domain performance residual and a first stability domain performance residual according to the linear nominal model and the identification model; the first frequency domain performance residual The difference is the frequency domain performance residual between the linear nominal model and the identification model, and the first stability domain performance residual is the stability domain performance residual between the linear nominal model and the identification model;

A fault detection module, configured to perform fault diagnosis according to the first time domain performance residual, the first frequency domain performance residual and the first stability domain performance residual.

Optionally, the fault detection module specifically includes:

a first normalization processing unit, configured to perform normalization processing on the first time domain performance residual, the first frequency domain performance residual and the first stability domain performance residual respectively;

The composite performance residual vector generating unit is used to form the normalized first time-domain performance residual, the normalized first frequency-domain performance residual, and the normalized first stable-domain performance residual. A point in the three-dimensional composite performance residual space; the coordinate system of the three-dimensional composite performance residual space takes the time domain performance residual as the x-axis, the frequency domain performance residual as the y-axis, and the stability domain performance residual as the z-axis, The point in the coordinate system and the origin form a composite performance residual vector;

The fault detection unit is used to judge the size of the composite performance residual vector length and the preset fault threshold; if the composite performance residual vector length is less than the preset fault threshold, the actuator of the actual control system has not failed; otherwise, the actuator of the actual control system malfunction.

Optionally, the closed-loop control system actuator fault diagnosis system further includes:

a characterization parameter obtaining module, used to obtain characterization parameters of different failure modes; the characterization parameters are parameters representing failures;

The linear fault model establishment module is used to establish the linear fault model of the control system model under the fault condition according to the characterization parameters of different fault modes;

a second output calculation module, configured to calculate the output of the linear fault model under closed-loop feedback according to the instruction input data;

a second time-domain performance residual module, configured to determine a second time-domain performance residual according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the linear fault model under closed-loop feedback; the The second time-domain performance residual is the time-domain performance residual between the linear nominal model and the linear fault model;

The second gap metric calculation module is configured to adopt the gap metric method according to the linear nominal model and the linear fault model to determine the second frequency domain performance residual and the second stability domain performance residual; the second frequency domain performance The residual is the frequency domain performance residual between the linear nominal model and the linear fault model, and the second stability domain performance residual is the stability domain performance residual between the linear nominal model and the linear fault model;

A fault diagnosis module, configured to judge the fault type and fault degree according to the second time domain performance residual, the second frequency domain performance residual and the second stability domain performance residual.

Optionally, the fault diagnosis module specifically includes:

A second normalization processing unit, configured to respectively normalize the second time domain performance residual, the second frequency domain performance residual and the second stability domain performance residual for each fault type unified treatment;

The actuator fault space library establishment unit is used for the normalized second time domain performance residual, the normalized second frequency domain performance residual and the normalized second stable performance according to each fault type Domain performance residuals to establish actuator fault space library;

The fault type judgment unit is used for judging the fault type and fault size in the actuator fault space library according to the direction of the composite performance residual vector.

Compared with the prior art, the beneficial effects of the present invention are:

The present invention proposes a performance-based fault diagnosis method and system for an actuator of a closed-loop control system. The time-domain performance residual between the nominal model and the identification model is determined according to the identification model of the actual control system and the nominal model of the control system model. , frequency domain performance residuals and stability domain performance residuals; fault diagnosis is carried out by establishing the connection between the actuator fault and the closed-loop system performance, which improves the reliability and practicability of fault diagnosis.

In addition, the present invention constructs a composite performance residual vector based on multi-domain (time domain, frequency domain and stability domain) information, which can comprehensively evaluate the actual control system performance after the occurrence of additive and multiplicative faults. The difference between the stability, rapidity and accuracy of the closed-loop control system and the nominal model, and the length information and direction information of the residual vector are used for fault diagnosis and fault identification respectively, avoiding single-domain or few-domain information in control. Inadequate system troubleshooting.

The invention also introduces the idea of gap measurement in robust control into the field of fault diagnosis, as the quantification basis of the rapidity residual and stability residual of the closed-loop control system to be diagnosed, and the performance residual based on the gap measurement will make the control system perform The failure of the mechanism is quantitatively related to the performance of the closed-loop system, which can effectively evaluate the closed-loop criticality of the failure, and provide a theoretical basis for the decision-making of the fault-tolerant control of the control system executive mechanism.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. In the embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort.

1 is a flowchart of a method for diagnosing faults of an executive mechanism of a closed-loop control system based on performance in an embodiment of the present invention;

2 is a schematic diagram of fault diagnosis based on multi-domain composite performance residuals in an embodiment of the present invention;

3 is a schematic diagram of the time domain-frequency domain-stability domain composite performance residual vector space in an embodiment of the present invention;

FIG. 4 is a structural diagram of a fault diagnosis system for an executive mechanism of a performance-based closed-loop control system in an embodiment of the present invention;

5 is a schematic diagram of a closed-loop control system for an electro-hydraulic actuator in an embodiment of the present invention;

Fig. 6 is the simulation result diagram of electro-hydraulic servo valve fault diagnosis in the embodiment of the present invention;

FIG. 7 is a vector diagram of the performance residuals of the electro-hydraulic servo valve in the embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The purpose of the present invention is to provide a performance-based method and system for diagnosing faults of an actuator of a closed-loop control system. By establishing the connection between the fault of the actuator and the performance changes of the control system after the closed-loop, the fault diagnosis is performed, and the reliability of the fault diagnosis of the control system is improved. properties and applicability.

In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

The present invention provides a performance-based fault diagnosis method for a closed-loop control system actuator. FIG. 1 is a flowchart of a performance-based closed-loop control system actuator fault diagnosis method in an embodiment of the present invention. As shown in FIG. 1 , the method includes:

Step 101: Acquire input data and output data of the actual control system.

Step 102: Establish a linear nominal model of the control system model under no fault condition according to the working principle of the control system.

y＝h(x,u)，其中，x为控制系统的状态向量，x＝[x₁,x₂,…,x_n]，u为控制系统的输入，y为控制系统的输出。As shown in Figure 2, an n-order nonlinear nominal model for describing a single-input single-output control system is established

y=h(x, u), where x is the state vector of the control system, x=[x ₁ , x ₂ ,..., x _n ], u is the input of the control system, and y is the output of the control system.

在第j个工作点

进行模型线性化处理：According to the Taylor expansion theorem, for higher-order nonlinear nominal models

at the jth working point

Perform model linearization:

表示第j个线性化的工作点，Δx表示在第j个工作点附近状态的微小增量，Δu表示在第j个工作点附近指令的微小增量。In the formula, g ^j (x, u) represents the rate of change of the state vector of the dynamic system, h ^j (x, u) represents the output of the dynamic system,

Represents the jth linearized operating point, Δx represents the small increment of the state near the jth operating point, and Δu represents the commanded small increment near the jth operating point.

in,

则得到所有工作点线性化标称模型：

Then the linearized nominal model for all operating points is obtained:

G ^j (s)=C ^j (sI-A ^j )- ¹ B ^j +D ^j

进行模型降阶处理。首先，采用下面公式计算系统矩阵A^j的Hankel奇异值：Using the Hankel singular value method for the linear nominal model G ^j (s) at the jth operating point

Perform model reduction. First, use the following formula to calculate the Hankel singular value of the system matrix A ^j :

Among them, λ _k , (k=1,2,...,n) is the k-th eigenvalue, P ^j , Q ^j are the controllability and objective character Ram arrays that satisfy the following relations:

A ^j P ^j +P ^j (A ^j ) ^T =-B ^j (B ^j ) ^T

(A ^j ) ^T Q ^j +Q ^j A ^j =-(C ^j ) ^T C ^j

According to the calculated Hankel singular values σ ^j | _k , keep the main singular values respectively, and obtain the reduced-order linear nominal model:

Step 103: Determine the identification model of the actual control system according to the input data and output data of the actual control system.

Under the step command input, the command output data of the controller and the measurement output data of the sensor in the actual system are collected, and then the collected data is filtered, cleaned and preprocessed, and the preprocessed data is saved for open-loop control of the actuator. Identification of the system model; using the least squares method to determine the parameters of the system model by minimizing the square sum function of the generalized error, and then obtain the identification model of the control system

Step 104: Obtain instruction input data.

Step 105: Calculate the output of the linear nominal model under the closed-loop feedback according to the instruction input data, and simultaneously calculate the output of the identification model under the closed-loop feedback according to the instruction input data.

Step 106: Determine a time-domain performance residual between the linear nominal model and the identification model according to the difference between the output of the linear nominal model under the closed-loop feedback and the output of the identification model under the closed-loop feedback.

在闭环反馈下的输出y_nominal(t)，计算实际运行系统辨识模型

在在同样的指令输入下闭环反馈的输出y_fault(t)，计算时域性能残差e_y：Computes a reduced-order linear nominal model with a step command input

Output y _nominal (t) under closed-loop feedback to calculate the actual operating system identification model

At the output y _fault (t) of the closed-loop feedback under the same command input, calculate the time-domain performance residual e _y :

e _y =y _fault (t)-y _nominal (t)

Step 107: According to the linear nominal model and the identification model, the gap metric method is used to determine the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the identification model.

和辨识得到的实际模型

之间的频域性能残差δ_ν：Calculation of Linear Nominal Model Using Gap Metric Formula

and the actual model identified

The frequency-domain performance residual between δ _ν :

When the feedback controller K makes the system G unstable, the stability is defined as the index of 0. When the feedback controller stabilizes the system, the following stability calculation method is defined:

和辨识得到的实际模型

之间的稳定域性能残差Δb_K：Calculate the linear nominal model using

and the actual model identified

The stability domain performance residual Δb _K between:

Step 108: Perform fault detection according to the time-domain performance residuals between the linear nominal model and the identification model, the frequency-domain performance residuals and the stability-domain performance residuals between the linear nominal model and the identification model.

Step 108 specifically includes:

The time domain performance residuals between the linear nominal model and the identification model, the frequency domain performance residuals and the stability domain performance residuals between the linear nominal model and the identification model are normalized respectively.

The normalized first time-domain performance residual, the normalized first frequency-domain performance residual, and the normalized first stable-domain performance residual are formed into points in the three-dimensional composite performance residual space. The coordinate system of the three-dimensional composite performance residual space takes the time domain performance residual as the x-axis, the frequency domain performance residual as the y-axis, and the stability domain performance residual as the z-axis. The points in the coordinate system and the origin form the composite performance residual. vector.

Determine the size of the composite performance residual vector length and the preset fault threshold; if the composite performance residual vector length is less than the preset fault threshold, the control system actuator does not fail; otherwise, the control system actuator fails.

The three performance residual indicators e _y , δ _ν and Δb _K are respectively normalized according to the following formulas:

e_s为控制系统稳态的容许误差，e_s≥0；α为最大故障程度因子。In the formula,

_es is the allowable error in the steady state of the control system, _es ≥ 0; α is the maximum failure degree factor.

In the formula, δ _t is the frequency domain index corresponding to the rapidity of the control system in the healthy state, 0≤δ _t ≤1, and β is the maximum failure degree factor.

为标称模型

的稳定性指标，

为实际模型

的稳定性指标，

γ为最大故障程度因子。In the formula,

for the nominal model

the stability index,

for the actual model

the stability index,

γ is the maximum failure degree factor.

The three normalized performance index residuals are formed into the following standardized composite performance residual vector H:

Step 109: Acquire characterization parameters of different failure modes. Characterization parameters are parameters that represent a fault.

Step 110: Establish a linear fault model of the control system model under fault conditions according to the characteristic parameters of different fault modes.

Combined with the working characteristics of the control system actuator, its typical failure modes are summarized, and the system parameters that can characterize the corresponding failure modes are determined based on the obtained high-order linear nominal model, and a high-order nonlinear fault model is obtained. By changing the system parameters that characterize the failure size for failure simulation. The difference between the nominal model and the fault model is that the parameter values that can characterize the fault type in the fault model are one in the normal value range and the other in the abnormal value range.

y＝h_ρ(x,u)，用于控制系统故障仿真获取数据。Further, combined with the working characteristics of the feedback control system actuator, summarize the typical failure mode set S _F = {F ₁ , F ₂ ,..., F _m }, and find the system parameters that can characterize these failure modes in the linear nominal model S _ρ ={ρ ₁ ,ρ ₂ ,...,ρ _m }, where ρ _i (i=1,2,...,m) represents the parameters of the system change affected by the ith fault ρ _i , and establishes a dynamic feedback control system The higher-order nonlinear fault model of

y=h _ρ (x, u), used to obtain data from control system fault simulation.

进行模型线性化处理：According to Taylor's expansion theorem, for higher-order nonlinear fault models

Perform model linearization:

in,

则得到所有工作点的线性故障模型：

Then the linear fault model of all operating points is obtained:

在第j个工作点

进行模型降阶处理。首先，采用下面公式计算

的Hankel奇异值：Linear fault model using Hankel singular value method

at the jth working point

Perform model reduction. First, use the following formula to calculate

The Hankel singular values of :

分别是满足如下关系的可控性和客观性格拉姆阵：Among them, λ _k , (k=1,2,...,n) is the kth eigenvalue,

They are the controllability and objective character Ram arrays that satisfy the following relations:

保留其中主要的奇异值，得到降阶的线性故障模型：According to the calculated Hankel singular values

Retaining the main singular values, a reduced-order linear fault model is obtained:

Step 111: Calculate the output of the linear fault model under closed-loop feedback according to the command input data.

Step 112: Determine a time domain performance residual between the linear nominal model and the linear fault model according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the linear fault model under closed-loop feedback.

Step 113: Determine the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the linear fault model by using the gap metric method according to the linear nominal model and the linear fault model.

Step 114: Judge the fault type and fault degree according to the time domain performance residual between the linear nominal model and the linear fault model, the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the linear fault model .

Step 114 specifically includes:

For each fault type, the time domain performance residuals between the linear nominal model and the linear fault model, the frequency domain performance residuals between the linear nominal model and the linear fault model, and the stability domain performance residuals are normalized separately. Unified processing.

The time domain performance residuals between the normalized linear nominal model and the linear fault model, the frequency domain performance residuals between the normalized linear nominal model and the linear fault model for each fault type, and The stability domain performance residuals between the normalized linear nominal model and the linear fault model are used to build the actuator fault space library.

The fault type is judged according to the direction of the composite performance residual vector in the actuator fault space library, and the fault degree is judged according to the length of the composite performance residual vector in the actuator fault space library. The vector corresponds to a point in the three-dimensional space, which are regarded as xyz coordinates respectively. The direction of the directed line between the healthy origin and this point is the direction of the vector, and the healthy origin is the origin when no fault occurs.

The actuator fault space library refers to simulating all possible faults of different types and sizes by means of tests or simulations, and then calculating the multi-domain performance residuals of the corresponding faults respectively. Due to the redundant performance of different faults, the residual vectors are located in different spatial regions in the three-dimensional coordinate system, that is, they have different directions of the residual vectors. Based on such a fault database established in advance, when a fault occurs in the actual system, its residual vector is calculated, and what type of fault has occurred is judged by the direction of the residual vector in the fault database. A fault space library is a collection of known directions or spatial distributions of various faults. The schematic diagram of the time domain-frequency domain-stability domain composite performance residual vector space is shown in Figure 3.

The present invention also provides a performance-based closed-loop control system actuator fault diagnosis system. FIG. 4 is a structural diagram of the performance-based closed-loop control system actuator fault diagnosis system in an embodiment of the present invention. As shown in FIG. 4 , the system includes:

The actual control system data acquisition module 201 is used for acquiring input data and output data of the actual control system.

The linear nominal model establishing module 202 establishes a linear nominal model of the control system model under no fault condition according to the working principle of the control system.

The identification model establishment module 203 is configured to determine the identification model of the actual control system according to the input data and output data of the actual control system.

The instruction input data acquisition module 204 is used for acquiring instruction input data.

The first output calculation module 205 is configured to calculate the output of the linear nominal model under the closed-loop feedback according to the instruction input data, and simultaneously calculate the output of the identification model under the closed-loop feedback according to the instruction input data.

The first time-domain performance residual determination module 206 is configured to determine the time-domain performance between the linear nominal model and the identification model according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the identification model under closed-loop feedback residual.

The first gap metric calculation module 207 is configured to adopt the gap metric method according to the linear nominal model and the identification model to determine the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the identified model.

The fault detection module 208 is configured to perform fault detection according to the time domain performance residual between the linear nominal model and the identification model, the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the identification model.

The fault detection module 208 specifically includes:

The first normalization processing unit is used to respectively perform the time domain performance residual between the linear nominal model and the identification model, the frequency domain performance residual between the linear nominal model and the identification model, and the stability domain performance residual. Normalized processing.

The composite performance residual vector generating unit is used to form the normalized first time-domain performance residual, the normalized first frequency-domain performance residual, and the normalized first stable-domain performance residual. A point in the 3D composite performance residual space. The coordinate system of the three-dimensional composite performance residual space takes the time domain performance residual as the x-axis, the frequency domain performance residual as the y-axis, and the stability domain performance residual as the z-axis. The points in the coordinate system and the origin form the composite performance residual. vector.

The fault diagnosis unit is used to judge the size of the composite performance residual vector length and the preset fault threshold; if the composite performance residual vector length is less than the preset fault threshold, the control system actuator is not faulty; otherwise, the control system actuator is faulty .

The characterization parameter acquisition module 209 is configured to acquire characterization parameters of different failure modes. Characterization parameters are parameters that represent a fault.

The linear fault model establishment module 210 is configured to establish a linear fault model of the control system model under a fault condition according to the characteristic parameters of different fault modes.

The second output calculation module 211 is configured to calculate the output of the linear fault model under closed-loop feedback according to the instruction input data.

The second time domain performance residual module 212 is configured to determine the time domain between the linear nominal model and the linear fault model according to the difference between the output of the linear nominal model under closed-loop feedback and the output of the linear fault model under closed-loop feedback performance residuals.

The second gap metric calculation module 213 is configured to use the gap metric method according to the linear nominal model and the linear fault model to determine the frequency domain performance residual and the stability domain performance residual between the linear nominal model and the linear fault model.

The fault diagnosis module 214 is used for classifying fault types according to the time domain performance residuals between the linear nominal model and the linear fault model, the frequency domain performance residuals and the stability domain performance residuals between the linear nominal model and the linear fault model and fault degree judgment.

The fault diagnosis module 214 specifically includes:

The second normalization processing unit is used for, for each fault type, the time domain performance residual between the linear nominal model and the linear fault model, and the frequency domain performance between the linear nominal model and the linear fault model, respectively. Residuals and stability domain performance residuals are normalized.

Actuator fault space library building unit for time-domain performance residuals between normalized linear nominal models and linear fault models, normalized linear nominal models and linear faults for each fault type The frequency domain performance residuals between the models and the stable domain performance residuals between the normalized linear nominal model and the linear fault model establish the actuator fault space library.

The fault type judgment unit is used to judge the fault type according to the direction of the composite performance residual vector in the actuator fault space library, and judge the fault degree according to the length of the composite performance residual vector in the actuator fault space library.

For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

其中，θ是伺服阀力矩马达的偏转角度，

是伺服阀力矩马达的偏转的角加速度，P₁是伺服阀阀芯左侧控制腔的压力，P₂是伺服阀阀芯右侧控制腔的压力，P₃是喷嘴后的压力，

是伺服阀芯的速度，x_s是伺服阀芯的位移，y是计量活门的输出位移。设力矩马达的控制输入电流为i_e，系统的供油压力为P_s。In the following, taking the actuator of a certain aviation equipment fuel control system as the diagnosis object, the effectiveness of the method of the present invention is verified by a simulation example. The fuel control system mainly includes electro-hydraulic servo valve, metering valve, differential pressure valve and other components. The electro-hydraulic servo valve and metering valve are the main components of the actuator, as shown in Figure 5. Since the displacement of the spool of the electro-hydraulic servo valve is usually very small, it is regarded as working near the zero working point, and the state vector is selected.

where θ is the deflection angle of the servo valve torque motor,

is the angular acceleration of the deflection of the servo valve torque motor, P ₁ is the pressure of the left control chamber of the servo valve spool, P ₂ is the pressure of the right control chamber of the servo valve spool, P ₃ is the pressure behind the nozzle,

is the speed of the servo spool, x _s is the displacement of the servo spool, and y is the output displacement of the metering valve. Let the control input current of the torque motor be i _e , and the oil supply pressure of the system be P _s .

Step 1. Establish a high-order nonlinear nominal mathematical model of the fuel control system actuator under no fault condition. Due to the large number of model parameters, the specific parameterized model is not given here, and the 8th-order nonlinear nominal model is obtained after directly substituting the specific values:

Step 2. Combined with the characteristics of the fuel control system actuator, its typical failure modes mainly include: torque motor electromagnetic performance degradation, nozzle pollution blockage, pilot stage oil filter pollution blockage, and feedback rod ball head gap wear fault.

The system parameters that characterize the above failure modes are: current-torque gain coefficient, effective diameter of nozzle, pilot stage control pressure and feedback rod ball head gap size. By changing the magnitude of these parameters in the simulation model, the degree of different faults is simulated, and the nonlinear fault model after the fault is obtained.

Step 3. Perform model linearization on the 8th-order nonlinear nominal model obtained in step 1 at the zero working point, and the results are as follows:

B ⁰ =[1.1×10 ⁶ 0 0 0 0 0 0 0] ^T

C ⁰ =[0 0 0 0 0 0 0 1], u= _ie

In the above formula, A ⁰ , B ⁰ , and C ⁰ are system matrices, and A ⁰ , B ⁰ , and C ⁰ are the cases of A ^j , B ^j , and C ^j when j = 0, which is a linear operation performed at the zero working point. , u is the input of the system.

The Hankel singular value method is used to reduce the order of the model and write it in the form of a transfer function to obtain a simplified open-loop nominal model of the fuel metering system:

Step 4. By means of simulation, the above four types of faults are injected into the high-order nonlinear model respectively, and the results are shown in Figure 6(a)-(d). Figure 6(a) is the simulation result of electromagnetic performance degradation. Fig. 6(b) is the simulation result of nozzle pollution and clogging, Fig. 6(c) is the simulation result of pilot oil filter clogging, and Fig. 6(d) is the simulation result of ball head gap wear. The input and output data of the system are collected and the system model is identified. The results are shown in Table 1.

Table 1 Electro-hydraulic servo control system faults and corresponding model identification

Step 5. Residual generation: According to the linear nominal model obtained in step 3 and the system output information measured in step 4, calculate the performance residual of the time domain index; calculate the simplified linear nominal model obtained in step 3 and step 4 according to the formula The frequency domain performance residuals and the stability domain performance residuals between the identified actual models are shown in Table 2.

Table 2 Multi-domain performance residuals of electro-hydraulic servo control system under different failure modes

Step 6. Determine the maximum failure degree factors α, β, γ corresponding to the time domain performance, frequency domain performance and stability domain performance according to the control requirement index of the fuel metering control system, and then calculate the three performance residuals obtained in step 5. After normalization, the composite performance residual vector under different faults is obtained as shown in Table 3.

)，稳定裕度≤45°(对应

)。设定稳态误差为0.2mm，上升时间为0.1s(对应δ_ν＝0.3336)，稳定域残差为0.8375时分别对应最严重故障的性能指标，对应α＝2,β＝0.4,γ＝0.7。According to the control quality requirements of a certain fuel control system, the steady-state error is less than or equal to 0.05mm, and the rise time is less than or equal to 0.04s (corresponding to

), stability margin≤45°(corresponding to

). When the steady-state error is set to 0.2mm, the rise time is 0.1s (corresponding to δ _ν = 0.3336), and the residual error of the stability domain is 0.8375, respectively corresponding to the performance indicators of the most serious faults, corresponding to α=2, β=0.4, γ=0.7 .

Table 3 Normalized standard multi-domain performance residuals

Step 7. Perform typical fault simulations on each operating point of the actuator by means of experiments or simulations, and repeat the methods in steps 5 and 6 to obtain residual vectors under different faults, and form a dynamic feedback control system actuator fault space library.

Step 8. According to the performance requirements of the control system, set the fault thresholds of the three performance residual components, according to . The length of the composite performance residual vector obtained in step 6 is used for fault diagnosis, and the direction of the composite performance residual vector and the fault space obtained in step 7 are used for fault isolation. As shown in Figure 7, the residual vectors of the four types of faults injected in this embodiment are:

H ₁ =[0.0000 0.3318 0.0000] ^T

H ₂ =[-1.0000 0.0001 0.0010] ^T

H ₃ =[0.0000 0.2359 0.2381] ^T

H ₄ =[−1.0000 0.8708 1.0000] ^T .

Among them, H ₁ is the residual vector of the electromagnetic performance degradation fault, H ₂ is the residual vector of the nozzle contamination and clogging fault, H ₃ is the residual vector of the pilot oil filter clogging fault, and H ₄ is the residual error of the ball head gap wear fault. vector.

In this paper, specific examples are used to illustrate the principles and implementations of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; meanwhile, for those skilled in the art, according to the present invention There will be changes in the specific implementation and application scope. In conclusion, the contents of this specification should not be construed as limiting the present invention.

Claims (8)

1. A performance-based fault diagnosis method for an actuating mechanism of a closed-loop control system is characterized by comprising the following steps:

acquiring input data and output data of an actual control system;

establishing a linear nominal model of a control system model under the condition of no fault according to the working principle of the control system;

determining an identification model of the actual control system according to the input data and the output data of the actual control system;

acquiring instruction input data;

calculating the output of the linear nominal model under closed-loop feedback according to the instruction input data, and calculating the output of the identification model under closed-loop feedback according to the instruction input data;

determining a first time domain performance residual error according to a difference value between the output of the linear nominal model under closed-loop feedback and the output of the identification model under closed-loop feedback; the first time domain performance residual error is a time domain performance residual error between the linear nominal model and the identification model;

determining a first frequency domain performance residual error and a first stable domain performance residual error by adopting a gap measurement method according to the linear nominal model and the identification model; the first frequency domain performance residual is a frequency domain performance residual between the linear nominal model and the identification model, and the first stable domain performance residual is a stable domain performance residual between the linear nominal model and the identification model;

and carrying out fault detection according to the first time domain performance residual error, the first frequency domain performance residual error and the first stable domain performance residual error.

2. The method of claim 1, wherein the performing fault detection based on the first time domain performance residual, the first frequency domain performance residual, and the first stability domain performance residual comprises:

respectively carrying out normalization processing on the first time domain performance residual error, the first frequency domain performance residual error and the first stable domain performance residual error;

forming points in a three-dimensional composite performance residual error space by the normalized first time domain performance residual error, the normalized first frequency domain performance residual error and the normalized first stable domain performance residual error; the coordinate system of the three-dimensional composite performance residual error space takes a time domain performance residual error as an x axis, a frequency domain performance residual error as a y axis and a stable domain performance residual error as a z axis, and points in the coordinate system and an origin point form a composite performance residual error vector;

judging the size of the composite performance residual vector length and a preset fault threshold value; if the length of the composite performance residual error vector is smaller than a preset fault threshold value, the actual control system executing mechanism does not have a fault; otherwise, the actual control system executing mechanism is in failure.

3. The method of claim 2, wherein the performing fault detection based on the first time domain performance residual, the first frequency domain performance residual, and the first stability domain performance residual further comprises:

acquiring characterization parameters of different fault modes; the characterization parameter is a parameter representing a fault;

respectively establishing linear fault models of the control system model under the fault condition according to the characterization parameters of different fault modes;

calculating the output of the linear fault model under closed-loop feedback according to the instruction input data;

determining a second time domain performance residual error according to a difference value between the output of the linear nominal model under closed-loop feedback and the output of the linear fault model under closed-loop feedback; the second time domain performance residual error is a time domain performance residual error between the linear nominal model and the linear fault model;

determining a second frequency domain performance residual error and a second stable domain performance residual error by adopting a gap measurement method according to the linear nominal model and the linear fault model; the second frequency domain performance residual is a frequency domain performance residual between the linear nominal model and the linear fault model, and the second stable domain performance residual is a stable domain performance residual between the linear nominal model and the linear fault model;

and judging the fault type and the fault degree according to the second time domain performance residual error, the second frequency domain performance residual error and the second stable domain performance residual error.

4. The method for diagnosing faults of an actuator of a performance-based closed-loop control system according to claim 3, wherein the determining of the fault type and the fault degree according to the second time-domain performance residual, the second frequency-domain performance residual and the second stable-domain performance residual specifically comprises:

respectively normalizing the second time domain performance residual error, the second frequency domain performance residual error and the second stable domain performance residual error aiming at each fault type;

establishing an executing mechanism fault space library according to the normalized second time domain performance residual error, the normalized second frequency domain performance residual error and the normalized second stable domain performance residual error of each fault type;

and judging the fault type and the fault size in the executing mechanism fault space library according to the direction of the composite performance residual error vector.

5. A performance-based closed loop control system actuator fault diagnostic system, comprising:

the actual control system data acquisition module is used for acquiring input data and output data of an actual control system;

the linear nominal model establishing module is used for establishing a linear nominal model of the control system model under the fault-free condition according to the working principle of the control system;

the identification model establishing module is used for determining an identification model of the actual control system according to the input data and the output data of the actual control system;

the instruction input data acquisition module is used for acquiring instruction input data;

the first output calculation module is used for calculating the output of the linear nominal model under closed-loop feedback according to the instruction input data and calculating the output of the identification model under closed-loop feedback according to the instruction input data;

the first time domain performance residual error determining module is used for determining a first time domain performance residual error according to a difference value between the output of the linear nominal model under the closed-loop feedback and the output of the identification model under the closed-loop feedback; the first time domain performance residual error is a time domain performance residual error between the linear nominal model and the identification model;

a first gap measurement calculation module, configured to determine a first frequency domain performance residual and a first stable domain performance residual by using a gap measurement method according to the linear nominal model and the identification model; the first frequency domain performance residual is a frequency domain performance residual between the linear nominal model and the identification model, and the first stable domain performance residual is a stable domain performance residual between the linear nominal model and the identification model;

and the fault detection module is used for carrying out fault detection according to the first time domain performance residual error, the first frequency domain performance residual error and the first stable domain performance residual error.

6. The system of claim 5, wherein the fault detection module comprises:

a first normalization processing unit, configured to perform normalization processing on the first time-domain performance residual, the first frequency-domain performance residual, and the first stable-domain performance residual, respectively;

a composite performance residual vector generating unit, configured to form the normalized first time domain performance residual, the normalized first frequency domain performance residual, and the normalized first stable domain performance residual into a point in a three-dimensional composite performance residual space; the coordinate system of the three-dimensional composite performance residual error space takes a time domain performance residual error as an x axis, a frequency domain performance residual error as a y axis and a stable domain performance residual error as a z axis, and points in the coordinate system and an origin point form a composite performance residual error vector;

the fault detection unit is used for judging the size of the composite performance residual error vector length and a preset fault threshold value; if the length of the composite performance residual error vector is smaller than a preset fault threshold value, the actual control system executing mechanism does not have a fault; otherwise, the actual control system executing mechanism is in failure.

7. The closed-loop control system actuator fault diagnostic system of claim 6, wherein the closed-loop control system actuator fault diagnostic system further comprises:

the characterization parameter acquisition module is used for acquiring characterization parameters of different fault modes; the characterization parameter is a parameter representing a fault;

the linear fault model establishing module is used for respectively establishing linear fault models of the control system model under the fault condition according to the characterization parameters of different fault modes;

the second output calculation module is used for calculating the output of the linear fault model under closed-loop feedback according to the instruction input data;

the second time domain performance residual error module is used for determining a second time domain performance residual error according to the difference value of the output of the linear nominal model under the closed-loop feedback and the output of the linear fault model under the closed-loop feedback; the second time domain performance residual error is a time domain performance residual error between the linear nominal model and the linear fault model;

the second gap measurement calculation module is used for determining a second frequency domain performance residual error and a second stable domain performance residual error by adopting a gap measurement method according to the linear nominal model and the linear fault model; the second frequency domain performance residual is a frequency domain performance residual between the linear nominal model and the linear fault model, and the second stable domain performance residual is a stable domain performance residual between the linear nominal model and the linear fault model;

and the fault diagnosis module is used for judging the fault type and the fault degree according to the second time domain performance residual error, the second frequency domain performance residual error and the second stable domain performance residual error.

8. The system of claim 7, wherein the fault diagnosis module comprises:

a second normalization processing unit, configured to, for each fault type, perform normalization processing on the second time-domain performance residual, the second frequency-domain performance residual, and the second stable-domain performance residual, respectively;

the executing mechanism fault space library establishing unit is used for establishing an executing mechanism fault space library according to the normalized second time domain performance residual error, the normalized second frequency domain performance residual error and the normalized second stable domain performance residual error of each fault type;

and the fault type judging unit is used for judging the fault type and the fault size in the executing mechanism fault space library according to the direction of the composite performance residual error vector.

Images