CN110989563A - Unmanned naval vessel fault estimation method based on iterative adaptive observer - Google Patents

Abstract

The invention relates to an unmanned naval vessel fault estimation method based on an iterative adaptive observer, belonging to the technical field of unmanned naval vessel control; decomposing an unmanned surface vessel model containing both failure of a steering engine and sensor faults into two subsystems through coordinate transformation, wherein the subsystem 1 only contains the steering engine faults, and the subsystem 2 only contains the sensor faults; aiming at the subsystem 1, designing a self-adaptive fault observer to estimate a steering engine efficiency factor; aiming at the subsystem 2, an iterative adaptive fault observer is designed to estimate the fault of the sensor; and establishing an error equation of the subsystem 1 and the subsystem 2, and judging the stability of the error system. The method can realize accurate estimation of the fault condition of the unmanned ship system, and give information such as the time of fault occurrence, the development process, the severity of the fault and the like, so that an operation center can conveniently monitor the safety of the unmanned ship; the method can estimate the failure condition of the steering engine and the failure of the sensor of the unmanned ship at the same time, thereby reducing the cost of fault-tolerant design.

Description

Unmanned naval vessel fault estimation method based on iterative adaptive observer

Technical Field

The invention relates to an unmanned naval vessel fault estimation method based on an iterative adaptive observer, and belongs to the technical field of unmanned naval vessel control.

Background

As the unmanned naval vessel is used as an autonomous motion platform working in a complex marine environment, the unmanned naval vessel has the characteristics of unmanned operation, intelligentization and the like, and the casualty problem can not be generated, so that the unmanned naval vessel can execute tasks in high-risk sea areas for a long time. Meanwhile, the unmanned naval vessel has a small and flexible structure and strong concealment, and can play an important role in the fields of navigation, information detection, marine resource exploration and the like by loading diversified equipment with different task requirements.

On the other hand, as the marine environment is complex and the climate condition is severe, when the unmanned surface vessel carries out scientific investigation and exploration tasks for a long time, various devices such as a steering engine and a sensor inevitably break down, so that the performance of the unmanned surface vessel system is reduced, and even high-precision detection equipment carried on the vessel body is damaged. Therefore, the amplitude and the frequency of the unmanned ship fault are visually displayed in real time through the fault estimation technology, the health condition of the unmanned ship is effectively monitored, and the method has important significance for improving the safety and the reliability of the unmanned ship system.

The problem of fault diagnosis and fault tolerance design of unmanned ships has been a research hotspot in the control field. However, the existing methods mainly have the following two problems: firstly, the existing surface boat fault diagnosis method usually carries out qualitative judgment on whether a fault occurs through a residual error evaluation function generated by a fault filter, but cannot provide accurate and quantitative information of the fault, such as fault amplitude, shape and the like, so that an operation center cannot judge the time, development process and fault severity of the unmanned boat system fault, and certain difficulty is brought to the compensation of subsequent faults and the design of fault tolerance strategies. Secondly, the existing unmanned ship fault diagnosis method usually assumes that only a steering engine fault exists on a naval ship or diagnoses the fault of a sensor. And the part with fault in the actual operation process is uncertain, so a plurality of diagnostic modules are often required to be designed on the unmanned ship, and the cost of fault-tolerant design is increased.

Disclosure of Invention

The invention aims to provide an unmanned ship fault estimation method based on an iterative adaptive observer, which aims to solve the problems that the existing unmanned ship fault diagnosis technology cannot provide quantitative information of faults and needs multiple groups of modules to monitor the faults of ships.

The purpose of the invention is realized as follows: an unmanned naval vessel fault estimation method based on an iterative adaptive observer specifically comprises the following steps:

step 1, decomposing an unmanned surface vessel model containing both failure of a steering engine and failure of a sensor into two subsystems through coordinate transformation, wherein the subsystem 1 only contains the failure of the steering engine, and the subsystem 2 only contains the failure of the sensor;

step 2, aiming at the subsystem 1 in the step one, designing an adaptive fault observer to estimate an efficiency factor of the steering engine;

step 3, aiming at the subsystem 2 in the step one, designing an iterative adaptive fault observer to estimate the fault of the sensor;

and 4, establishing an error equation of the subsystem 1 and the subsystem 2, and judging the stability of the error system.

The invention also includes such structural features:

1. the step 1 specifically comprises the following steps:

step 1.1, establishing an unmanned surface vessel mathematical model with steering engine failure and sensor faults:

y_o(t)＝C_ox_o(t)+D_osf_s(t)

wherein the state vector x_o(t)＝[v(t)r(t)ψ(t)p(t)φ(t)]^TInterference vector d (t) ═ w_ψw_φ]^TControl input u (t) is δ (t), ρ (t) is an unknown actuator efficiency factor, y_o(t)∈R^pIs to measure the output signal, f_s(t)＝[f_s1,f_s2,...,f_sq]^T∈R^qIs a sensor fault vector; variables v (t), ψ (t), φ (t), r (t) and p (t) represent the rudder generated unmanned boat sideslip velocity, heading angle, roll angle, yaw velocity and roll velocity, respectively; δ (t) represents a rudder angle; w is a_φ(t) and w_ψ(t) represents the disturbance of the roll angle and course angle caused by sea waves; t is_v and T_rIs a time constant; k_vr，K_dv，K_dv，K_dr，K_dp，K_vpIs the known gain; w is a_nAnd ζ represent the undamped natural frequency and the damping ratio, respectively, and the system matrix is represented as:

Co＝I₅

other model parameters, disturbances and faults need to be satisfied: d_os∈R^5×qA sensor fault coefficient matrix is obtained, and q is less than or equal to 5;

the efficiency factor rho (t) of the steering engine meets the condition that rho (t) is more than 0 and less than or equal to rho_uLess than or equal to 1; interference d (t) satisfies | | d (t) | | ≦ d; sensor failure f_sv(t) satisfies | f_sv(t)|≤s_uv，

d*，s_uv，s_dvIs an unknown positive number;

step 1.2, for the original system 1, solving the reversible matrix M e R^5×5 and N∈R^5×5And satisfies the following conditions:

wherein ,A₁∈R^3×3，G₁∈R^3×1，G₂∈R^3×2，C₁∈R^3×3Is reversible, D ∈ R^2×q；

Step 1.3, the linear transformation x (t) ═ Mx is introduced_o(t)，y(t)＝Ny_o(t)，

x₁(t)∈R³，y₁(t)∈R³And according to the formula in the step 1.2, two subsystems which only contain the faults of the steering engine and the sensor can be obtained respectively:

step 1.4, the process of carrying out the augmentation treatment on the sensor fault formula in the step 1.3 comprises the following steps:

first, a measurable output variable is defined

Then, an augmented vector is constructed

Given a

Then the rudder fault formula in step 1.4 can be rewritten as:

subsystem 1:

finally, according to the sensor failure formula in step 1.3 and the first formula in step 1.4, the augmented subsystem 2 can be obtained:

and (3) subsystem 2:

wherein ξ (t) ∈ R²Is the measurement output of the subsystem 2.

2. The step 2 specifically comprises: for subsystem 1 in the subsystem formula in step 1.4, an adaptive observer is designed as follows:

wherein ,

is x₁(t) an estimate of;

is that

Estimate in the kth iteration observer;

is an estimate of the actuator efficiency factor ρ (t); a. the_f∈R^3×3Is the matrix parameter to be designed, will be at step 4.2 matrix A_fSolving the formula to give; observer input u_d(t) is designed as:

wherein ,

is d^*Adaptive estimation of (2); p is belonged to R^3×3Is a positive definite symmetric matrix to be designed, which will be given in the first formula of step 4.2;

b₀ and b₁Is a given positive number;

and

the adaptive law of (c) is designed as:

wherein ,c₁，c_d1，c_d2Is a given positive number.

3. The step 3 specifically includes: for the subsystem 2 in the subsystem 2 formula in step 1.4, an iterative adaptive observer is designed as follows:

wherein ,

is that

The k-th iteration observer of (1). θ is the maximum number of iterations. L is_p∈R^4×2Is that the observer gain is satisfied with

Is a stable matrix;

is an estimated value of sensor fault, and the iteration self-adaptive rate is as follows: when k is 1:

when k is more than or equal to 2:

wherein ,c_svIs a given positive number; II type_1vIs a matrix pi₁∈R^q×2V th row of (ii)_2vIs a matrix pi₂∈R^q×2Row v 1,2,.., q, matrix Π₁ and Π₂Calculated by the first formula of step 4.2.

4. The step 4 of judging the stability of the error system specifically comprises the following steps:

step 4.1, defining the estimation error as:

establishing an error system dynamic equation as follows:

and 4.2, judging the stability of the error upper formula according to the following conditions:

if there is a positive value constant δ_sγ, positive definite matrix P ∈ R^3×3，Q∈R^4×4Matrix of

Π₁∈R^q×2，Π₂∈R^q×2And satisfies the following conditions:

wherein

The dynamic equation of the error system is bounded and stable, and the parameter matrix P belongs to R^3×3N, matrix II₁∈R^q×2，Π₂∈R^q×2Can be obtained by solving the first formula in step 4.2, and the matrix A_fCan be solved as:

other adaptive parameters c in the formula of step 2, step 3 detailed steps₁,c_d1,c_d2,c_s1,b₀,b₁Given a free positive number, adjustments can be made according to observer effects.

Compared with the prior art, the invention has the beneficial effects that: the unmanned ship fault monitoring system can provide quantitative information such as amplitude, shape and the like of the unmanned ship fault in real time, visually display the development process of the fault and the severity of the fault, facilitate the monitoring of the unmanned ship safety by the control center, and improve the reliability of the unmanned ship in the task execution process. The invention can also monitor the failure condition of the steering engine of the unmanned ship and the fault of the sensor in a unified way, reduces the construction of a fault diagnosis module on the unmanned ship and effectively reduces the cost of fault-tolerant design.

Drawings

FIG. 1 is a flow chart of the present invention;

FIG. 2 is a system block diagram of the present invention;

FIG. 3 is a schematic diagram of a steering engine real efficiency factor (solid line) and a steering engine efficiency factor estimate (dashed line) in a simulation example;

FIG. 4 is a schematic diagram of a true sensor fault (solid line) versus an estimated value of the sensor fault (dashed line) in a simulation example.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

The invention provides an unmanned surface vessel fault estimation method based on an iterative adaptive observer, which comprises the following specific implementation steps as shown in figure 1:

step one, through coordinate transformation with contain the unmanned surface of water ship model of steering wheel inefficacy, sensor trouble simultaneously and decompose into two subsystems, wherein subsystem 1 only contains the steering wheel trouble, and subsystem 2 only contains the sensor trouble, step one includes:

step A, according to a document 'Integrated-Based Event-Triggered Fault Detection Filter design for Unmanned Surface Vehicles', establishing a mathematical model of the Unmanned Surface vehicle with the failure of a steering engine and the failure of a sensor as follows:

y_o(t)＝C_ox_o(t)+D_osf_s(t) (1)

wherein the state vector x_o(t)＝[v(t)r(t)ψ(t)p(t)φ(t)]^TInterference vector d (t) ═ w_ψw_φ]^TControl input u (t) is δ (t), ρ (t) is an unknown actuator efficiency factor, y_o(t)∈R^pIs to measure the output signal, f_s(t)＝[f_s1,f_s2,...,f_sq]^T∈R^qIs a sensor fault vector; variables v (t), ψ (t), φ (t), r (t) and p (t) represent the rudder generated unmanned boat sideslip velocity, heading angle, roll angle, yaw velocity and roll velocity, respectively; δ (t) represents a rudder angle; w is a_φ(t) and w_ψ(t) represents the disturbance of the roll angle and course angle caused by sea waves; t is_v and T_rIs a time constant; k_vr，K_dv，K_dv，K_dr，K_dp，K_vpIs the known gain; w is a_nAnd ζ represent the undamped natural frequency and the damping ratio, respectively, and the system matrix is represented as:

the parameters are given as follows: k_vr＝-0.46m/s,ω_n＝0.63rad/s,ζ＝0.0936，K_dr＝-0.0211,T_r＝1.6667,T_v＝10,K_dv＝0.0780,K_vp＝1.6380，K_dp-0.0852, a parameter matrix can be obtained:

B_o＝[0.0078 -0.0126 0 -0.03380]^T,

C_o＝I₅

the sensor failure coefficient matrix is assumed to be D_os＝[0 0 1 0 1]^T。

Step B, solving the invertible matrix for the original system (1)

And

satisfies the following conditions:

step C, according to equation (3), introducing a linear transformation x (t) ═ Mx_o(t)，y(t)＝Ny_o(t), wherein

x₁(t)∈R³，y₁(t)∈R³Two subsystems containing only steering engine faults and sensor faults respectively can be expressed as:

step D, the process of carrying out augmentation treatment on the sensor fault in the formula (5) is as follows:

first, a measurable output variable is defined

Then, an augmented vector is constructed

Given a

Equation (4) can be rewritten as:

subsystem 1:

finally, from equations (5) and (6), an augmented subsystem 2 is obtained:

and (3) subsystem 2:

wherein ξ (t) ∈ R²Is the measurement output of the subsystem 2.

Step two, aiming at the subsystem 1 in the step one, designing an adaptive fault observer to estimate a steering engine efficiency factor, wherein the step two comprises the following steps:

step E, designing an adaptive observer for the subsystem 1 in the formula (7) as follows:

wherein ,

is x₁(t) an estimate of;

is that

Estimate in the kth iteration observer;

is an estimate of the actuator efficiency factor ρ (t); a. the_f∈R^3×3Is a matrix parameter to be designed, and will be given in equation (20); observer input u_d(t) is designed as:

wherein ,

is d^*Adaptive estimation of (2); p is belonged to R^3×3Is the positive definite symmetric matrix to be designed, which will be given in (18);

b₀ and b₁Is a given positive number.

And

the adaptive law design of (1) is as follows:

wherein ,c₁，c_d1，c_d2Is a given positive number.

Step three, aiming at the subsystem 2 in the step one, designing an iterative adaptive fault observer to estimate the fault of the sensor, wherein the step three comprises the following steps:

step F, designing an iterative adaptive observer for the subsystem 2 in the formula (8) as follows:

wherein ,

is that

The k-th iteration observer of (1). θ is the maximum number of iterations.

Is that the observer gain is satisfied with

Is a stable matrix;

is an estimated value of sensor fault, and the iteration self-adaptive rate is as follows: when k is 1:

when k is more than or equal to 2:

wherein ,c_svIs a positive number; II type_1vIs a matrix pi₁∈R^q×2V th row of (ii)_2vIs a matrix pi₂∈R^q×2Row v 1,2,.., q, matrix Π₁ and Π₂Calculated by equation (18).

Step four, establishing an error equation of the subsystem 1 and the subsystem 2, and giving a stability condition of an error system, wherein the step four comprises the following steps:

first, the estimation error is defined as:

establishing an error system dynamic equation as follows:

the stability of error equation (17) is judged according to the following conditions:

if there is a positive value constant δ_sγ, positive definite matrix P ∈ R^3×3，Q∈R^4×4Matrix of

Π₁∈R^q×2，Π₂∈R^q×2Is satisfied with

wherein

The dynamic error (17) is bounded and stable, and the parameter matrix P belongs to R^3×3，Π₁∈R^q×2，Π₂∈R^q×2Can be obtained by solving (18), each parameter matrix is solved as:

the other adaptive parameters in the formulae (10) to (12), (14) to (15) are given as c₁＝1000,c_d1＝1,c_d2＝1,c_s1＝2.1,b₀＝0.01，b₁＝0.01。

By the above design process, we are based on the adaptive observer (9) and the update rate (11)The estimated value of the efficiency factor rho (t) of the steering engine can be obtained

Meanwhile, according to the iterative adaptive observer (13) and the iterative update rates (14) - (15), the sensor fault f of the kth iterative process can be obtained_s(t) estimated value

To verify the effect of the present invention, the following simulation example was used for verification. Assuming that the failure condition of the steering engine and the sensor fault are as follows:

ρ(t)＝0.2e^-0.2t+0.35

disturbance w of course angle caused by sea waves_ψ(t) is [ -1,1]Random value of (1), roll angle disturbance w caused by sea waves_φ(t) 0.5sin (1.2t), and control input u (t) 0.1sin (t).

The steering engine failure condition and the observer estimation result are shown in fig. 3, where the solid line is a real curve of the steering engine efficiency factor, and the dotted line is the steering engine efficiency factor estimation value obtained by the observer. The sensor fault and observer estimation results are shown in fig. 4, where the solid line is the true curve of the sensor fault and the dashed line is the estimated value of the sensor fault obtained by the observer.

According to the method, the efficiency factor of the unmanned surface boat rudder machine and the sensor fault can be estimated simultaneously, the occurrence condition of the fault can be described quantitatively rapidly and accurately, and the safety of the unmanned boat can be monitored by a control center conveniently.

Descriptions not related to the embodiments of the present invention are well known in the art, and may be implemented by referring to the well-known techniques. The present invention is not limited to the above embodiments, and any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention should be covered in the protection scope of the present invention.

In summary, the invention discloses an unmanned surface vessel fault estimation method based on an iterative adaptive observer, which comprises the following steps: step one, decomposing an unmanned surface vessel model containing failure of a steering engine and failure of a sensor into two subsystems; designing a self-adaptive fault observer to estimate a steering engine efficiency factor; designing an iterative adaptive fault observer to estimate the fault of the sensor; and step four, judging the stability of the error system. The method can realize accurate estimation of the fault condition of the unmanned ship system, and give information such as the time of fault occurrence, the development process, the severity of the fault and the like, so that an operation center can conveniently monitor the safety of the unmanned ship; the method can estimate the failure condition of the steering engine and the failure of the sensor of the unmanned ship at the same time, avoids building a plurality of failure diagnosis modules on the unmanned ship, and reduces the cost of fault-tolerant design.

Claims (1)

1. An unmanned naval vessel fault estimation method based on an iterative adaptive observer is characterized by comprising the following steps:

step 1: decomposing an unmanned surface vessel model containing both failure of a steering engine and sensor faults into two subsystems through coordinate transformation, wherein the subsystem 1 only contains the steering engine faults, and the subsystem 2 only contains the sensor faults;

step 1.1: establishing an unmanned surface vessel mathematical model with steering engine failure and sensor faults:

y_o(t)＝C_ox_o(t)+D_osf_s(t)

wherein the state vector x_o(t)＝[v(t) r(t) ψ(t) p(t) φ(t)]^TInterference vector d (t) ═ w_ψw_φ]^TControl input u (t) is δ (t), ρ (t) is an unknown actuator efficiency factor, y_o(t)∈R^pIs to measure the output signal, f_s(t)＝[f_s1,f_s2,...,f_sq]^T∈R^qIs a sensor fault vector; variables v (t), ψ (t), φ (t), r (t) and p (t) represent the rudder generated unmanned boat sideslip velocity, heading angle, roll angle, yaw velocity and roll velocity, respectively; δ (t) represents a rudder angle; w is a_φ(t) and w_ψ(t) represents the disturbance of the roll angle and course angle caused by sea waves; t is_v and T_rIs a time constant; k_vr，K_dv，K_dv，K_dr，K_dp，K_vpIs the known gain; w is a_nAnd ζ represent the undamped natural frequency and the damping ratio, respectively, and the system matrix is represented as:

C_o＝I₅

other model parameters: interference and faults need to be satisfied: d_os∈R^5×qA sensor fault coefficient matrix is obtained, and q is less than or equal to 5;

the efficiency factor rho (t) of the steering engine meets the condition that rho (t) is more than 0 and less than or equal to rho_uLess than or equal to 1; the interference d (t) satisfies | | d (t) | | < d ≦ d^*(ii) a Sensor failure f_sv(t) satisfies | f_sv(t)|≤s_uv，

v＝1,2,...,q；d^*，s_uv，s_dvIs an unknown positive number;

step 1.2: for the original system 1, solving the invertible matrix M E R^5×5 and N∈R^5×5And satisfies the following conditions:

wherein ,A₁∈R^3×3，G₁∈R^3×1，G₂∈R^3×2，C₁∈R^3×3Is reversible, D ∈ R^2×q；

Step 1.3: introducing a linear transformation x (t) ═ Mx_o(t)，y(t)＝Ny_o(t)，

x₁(t)∈R³，y₁(t)∈R³Obtaining two subsystems only containing the faults of the steering engine and the sensor:

step 1.4: carrying out augmentation processing on a sensor fault formula:

first, a measurable output variable is defined

Then, an augmented vector is constructed

Given a

The steering engine fault formula can be rewritten as:

subsystem 1:

finally, the augmented subsystem 2 is obtained:

and (3) subsystem 2:

wherein ξ (t) ∈ R²Is the measurement output of subsystem 2;

step 2: for the subsystem 1, estimating a steering engine efficiency factor according to the adaptive fault observer;

wherein ,

is x₁(t) an estimate of;

is that

Estimate in the kth iteration observer;

is an estimate of the actuator efficiency factor ρ (t); a. the_f∈R^3×3Is the matrix parameter to be calculated; observer input u_d(t) is:

wherein ,

is d^*Adaptive estimation of (2); p is belonged to R^3×3Is a positive definite symmetric matrix to be calculated;

b₀ and b₁Is a given positive number;

and

the adaptive law of (1) is as follows:

wherein ,c₁，c_d1，c_d2Is a given positive number;

and step 3: for the subsystem 2, estimating the sensor fault according to an iterative adaptive fault observer;

wherein ,

is that

The estimated value in the kth iteration observer of (1); θ is the maximum number of iterations; l is_p∈R^{4 ×2}Is that the observer gain is satisfied with

Is a stable matrix;

is an estimated value of sensor fault, and the iteration self-adaptive rate is as follows:

when k is 1:

when k is more than or equal to 2:

wherein ,c_svIs a given positive number; II type_1vIs a matrix pi₁∈R^q×2V th row of (ii)_2vIs a matrix pi₂∈R^q×2Line v 1, 2.., q;

and 4, step 4: establishing an error equation of the subsystem 1 and the subsystem 2, and judging the stability of an error system;

step 4.1: defining the estimation error as:

establishing an error system dynamic equation as follows:

step 4.2: the stability of the error system is judged according to the following conditions:

if there is a positive value constant δ_sγ, positive definite matrix P ∈ R^3×3，Q∈R^4×4Matrix of

Π₁∈R^q×2，Π₂∈R^{q ×2}And satisfies the following conditions:

wherein

The dynamic equation of the error system is bounded and stable, and the parameter matrix P belongs to R^3×3Matrix A_fCan be solved as:

Images