A Method for Suppressing Harmonic Interference and Active Fault Tolerance of the Attitude System of a Flying Wing UAV

Abstract

The invention discloses a method for suppressing harmonic interference and actively tolerating faults in the attitude system of a flying wing unmanned aerial vehicle. First, a perturbed dynamics model of the attitude system of the flying wing unmanned aerial vehicle affected by rudder surface faults and harmonic interference is established; secondly, roll, pitch, and yaw three-channel observers are designed based on harmonic observer technology to realize the estimation of lumped interference; then, based on the lumped interference estimation information, a composite nonlinear dynamic inverse controller is reconstructed on the basis of the benchmark nonlinear dynamic inverse control algorithm; finally, based on the pseudo-inverse method control allocation technology, the desired torque is converted into the rudder surface deflection of the flying wing unmanned aerial vehicle. The invention regards the influence of rudder surface faults as interference, then makes full use of the interference frequency information, designs a harmonic interference observer to realize the high-precision estimation of lumped interference, and realizes the effective compensation of interference in the form of interference estimation feedforward, suppresses the influence of multi-source harmonic interference and faults on the control performance of the flying wing unmanned aerial vehicle, and improves the flight performance of the flying wing unmanned aerial vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of flight control, and particularly relates to a method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system. Background Technique

[0002] Due to its unique aerodynamic layout, the flying wing UAV has advantages such as large lift, strong maneuverability, and good stealth ability, and has been widely used in both military and civilian fields. Due to its characteristics such as small weight and no vertical tail, the flying wing UAV has characteristics such as strong coupling and poor maneuverability, and faces many challenges in controller design. Because of its high flexibility and strong maneuverability, the flying wing UAV often performs tasks in harsh and complex environments, and the frequent use of actuators will also lead to a high failure rate of actuator changes; therefore, the flying wing UAV faces the influence of multi-source interference and faults such as external environmental interference, internal aerodynamic parameter perturbation, model uncertainty, and actuator faults; these complex multi-source interferences and actuator faults will further exacerbate the non-linear characteristics of the UAV attitude system, bringing great challenges to controller design.

[0003] Regarding the control problem of the flying wing UAV attitude system under the action of multi-source interference and actuator faults, domestic and foreign scholars have proposed a variety of solutions, including dynamic inverse control based on a non-linear nominal model, sliding mode variable structure control, and composite control algorithms based on extended state observer technology. However, these methods passively eliminate the influence of interference based on the error signal or feed forward to compensate for the influence of the interference signal based on linear observer technology, and cannot adapt to high-frequency, fast time-varying, and highly complex interference environments. Therefore, there is an urgent need to propose a control method for the flying wing UAV attitude system that can quickly and accurately suppress the influence of complex interference and faults.

[0004] In addition, since the flying wing UAV adopts a redundant rudder surface configuration to ensure the overall reliability of the system, the available rudder surface is usually larger than the required rudder surface. Therefore, how to ensure the minimum required rudder surface drive energy without affecting the flight performance of the UAV is also an issue worthy of attention. Summary of the Invention

[0005] Object of the Invention: The present invention provides a method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system, which can realize the estimation of the lumped interference caused by harmonic interference and the influence of actuator faults, and feedback the interference estimation information to the controller in the form of feed forward to compensate for the adverse effects brought by multi-source interference and faults, so as to ensure that the flying wing UAV has stronger anti-interference performance and fault tolerance ability; and ensure that the rudder surface drive energy consumed while achieving the desired torque is minimized.

[0006] Technical Solution: The present invention provides a method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system, including the following steps:

[0007] (1) Establish a perturbed dynamics model of the flying wing UAV attitude system affected by control surface faults and harmonic disturbances, and obtain the dynamic tracking error of the flying wing UAV attitude system;

[0008] (2) Design roll, pitch, and yaw three-channel observers for the perturbed attitude system of the flying wing UAV based on harmonic observer technology;

[0009] (3) For the dynamic tracking error of the flying wing UAV attitude system, combine the interference estimation information of the harmonic observer to construct a composite nonlinear dynamic inverse attitude controller for the flying wing UAV, and obtain the virtual control quantity;

[0010] (4) Based on the pseudo-inverse method, convert the desired torque into the control surface deflection of the flying wing UAV.

[0011] Furthermore, the perturbed dynamics model of the flying wing UAV attitude system described in step (1) is:

[0012]

[0013] where s φ , t θ , c φ and c θ represent sinφ, tanθ, cosφ, and cosθ respectively; φ, θ, and ψ represent the roll angle, pitch angle, and yaw angle of the flying wing UAV respectively, and represent the first derivatives of φ, θ, and ψ respectively; p, q, and r represent the rotational angular velocities of the flying wing UAV around the body x, y, and z axes respectively, and represent the first derivatives of p, q, and r respectively; I x , I y and I z represent the moments of inertia of the flying wing UAV around the body x, y, and z axes respectively; I xz represents the product of inertia of the flying wing UAV; τ x , τ y and τ z represent the aerodynamic torques of the flying wing UAV around the body x, y, and z axes respectively; D x , D y and D z represent the periodic disturbances acting on the body axis, and their dynamics satisfy:

[0014]

[0015] where ω x , ω y , ω z represent the frequencies of the three-axis disturbances, and their values are known; Indicates the internal dynamics of interference Respectively represent and The first-order derivative of;

[0016] Through the moment in the disturbed dynamics model of the flying-wing UAV attitude system, the flying-wing UAV rudder loop model including control surface faults is obtained:

[0017] Γ = B[(I - K f )δ n +δ f = Bδ n +B(δ f -K f δ n ) = Γ n +Γ f (3)

[0018] Where Γ = [τ x τ y τ z T Is the moment vector; Represents the nominal moment vector; Represents the moment vector caused by the influence of control surface faults; B ∈ R 3×8 Represents the control efficiency matrix of the flying-wing UAV control surface; I ∈ R 8×8 Is the identity matrix; K f = diag{k f1 ,k f2 ,…,k f8} Represents the fault coefficient matrix of 8 control surfaces; δ n = [δ n1 δ n2 δ n3 δ n4 δ n5 δ n6 δ n7 δ n8 T Represents the nominal deflection angle value of the flying-wing UAV control surface ignoring the influence of faults, δ f = diag{δ f1 ,δ f2 ,…,δ f8} Represents the control surface jamming angle; Γ n Represents the nominal aerodynamic moment without control surface faults, Γ f Represents the aerodynamic moment caused by faults; The following definitions are introduced:

[0019]

[0020] Where, △ = I x I z -I​​xz 2 Considering the above definitions, equation (3) and equation (1) can be rewritten as follows:

[0021]

[0022] in is the first-order derivative of Θ, is the first-order derivative of Ω, represents the aggregate interference, and the expression is:

[0023] D A =D+GΓ f

[0024] According to the attitude system dynamics (4), the second-order dynamics of the attitude angle can be obtained as:

[0025] · in is the first-order derivative of W, D L is the aggregate interference including multi-source interference and control surface failure, and its expression is:

[0026]

[0027] Define the flying wing drone attitude tracking error:

[0028]

[0029] where Θ d =[φ d θ d ψ d ] T ,φ d ,θ d ,ψ d are the expected attitude angle command, e φ ,e θ ,e ψ Respectively represent the rolling, pitching, and yaw angle tracking errors, and the attitude system tracking error dynamics is:

[0030]

[0031] in and Respectively represent e Θ The second and first derivatives of and Respectively represent Θ d The second and first derivatives of .

[0032] Furthermore, the implementation process of step (2) is as follows:

[0033]

[0034] Among them, is the dynamics of the harmonic observer, and K ω = diag{ω x , ω y , ω z} represents the frequency matrix of harmonic interference; are the derivatives of Z1, Z2, and Z3 respectively, is the estimated value of D A , and L1, L2, and L3 are the gains of the harmonic observer. Their specific forms are as follows:

[0035]

[0036] Furthermore, the step (3) is implemented by the following formula:

[0037]

[0038] Among them, is the estimated information of the multi-source interference D A , and are the controller parameters. Their specific forms are as follows:

[0039]

[0040] Among them, and are control parameters and are both positive constants.

[0041] Furthermore, the implementation process of the step (4) is as follows:

[0042] According to the virtual control quantity Γ n obtained in step (3), based on the rudder loop equation, the true control quantity δ n of the flying wing UAV attitude system is obtained by inverse solution based on the pseudo-inverse method:

[0043] δ n = B + Γ n

[0044] Among them, B + is the generalized inverse matrix of B.

[0045] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention uses a harmonic observer to estimate complex disturbances and control surface faults in the attitude system, achieving the asymptotic estimation of lumped disturbances; the lumped disturbance estimation information is fed back into the design of the nonlinear dynamic inversion controller and reconstructed into a composite dynamic inversion controller. By dynamically and real-time feed-forward compensating the lumped disturbances, the flight performance of the UAV under disturbances and faults is ensured; the desired torque of the controller is solved by the pseudo-inverse method, ensuring that while the flying wing UAV accurately tracks the commands, the overall deflection angle of the control surfaces of the flying wing UAV is minimized, saving system energy. Description of the Drawings

[0046] Figure 1 is the flowchart of the present invention;

[0047] Figure 2 are the curves of the attitude angle tracking errors of the roll, pitch, and yaw channels of the flying wing UAV under the actions of the methods CNDIC+HO, CNDIC+ESO, and NDCI proposed by the present invention respectively;

[0048] Figure 3 are the curves of the control quantity responses of the roll, pitch, and yaw channels of the flying wing UAV under the actions of the methods CNDIC+HO, CNDIC+ESO, and NDCI proposed by the present invention respectively;

[0049] Figure 4 are the curves of the deflection angle responses of the control surfaces 1-4 of the flying wing UAV under the actions of the methods CNDIC+HO, CNDIC+ESO, and NDCI proposed by the present invention respectively;

[0050] Figure 5 are the curves of the deflection angle responses of the control surfaces 5-8 of the flying wing UAV under the actions of the methods CNDIC+HO, CNDIC+ESO, and NDCI proposed by the present invention respectively;

[0051] Figure 6 are the curves of the lumped disturbance estimation error responses under the actions of the methods HO and ESO proposed by the present invention. Detailed Implementation Manner

[0052] The present invention will be further described in detail below with reference to the drawings.

[0053] The present invention provides a method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system, as Figure 1 shown, which specifically includes the following steps:

[0054] Step 1: Establish a perturbed dynamic model of the flying wing UAV attitude system affected by control surface faults and harmonic interference.

[0055] Perturbed dynamic model of the flying wing UAV attitude system:

[0056]

[0057] where s φ , t θ , c φ and c θ represent sinφ, tanθ, cosφ, and cosθ respectively; φ, θ, and ψ represent the roll angle, pitch angle, and yaw angle of the flying-wing UAV respectively, represent the first-order derivatives of φ, θ, and ψ respectively; p, q, and r represent the rotational angular velocities of the flying-wing UAV about the body x, y, and z axes respectively, and represent the first-order derivatives of p, q, and r respectively; I x , I y and I z represent the moments of inertia of the flying-wing UAV about the body x, y, and z axes respectively; I xz represents the product of inertia of the flying-wing UAV; τ x , τ y and τ z represent the aerodynamic torques of the flying-wing UAV about the body x, y, and z axes respectively; D x , D y and D z represent the periodic disturbances acting on the body axes, and their dynamics satisfy:

[0058]

[0059] where ω x , ω y , ω z represent the frequencies of the three-axis disturbances, and their values are known; represents the disturbance internal dynamics, represent the first-order derivatives of respectively. In this example, ω x = 2, ω y = 3, ω z = -2.

[0060] Through the torques in the disturbed dynamics model of the flying-wing UAV attitude system, the flying-wing UAV rudder loop model including the rudder surface fault is obtained:

[0061] Γ = B[(I - K f )δ n + δ f = Bδ n + B(δ f - K f δ n ) = Γ n + Γ f , (3)

[0062] where Γ = [τ x τ y τ z T is the torque vector; represents the nominal torque vector; represents the torque vector affected by the control surface failure; B ∈ R 3×8 represents the control efficiency matrix of the control surfaces of the flying wing UAV; I ∈ R 8 ×8 is the identity matrix; K f = diag{k f1 , k f2 , …, k f8} represents the fault coefficient matrix of 8 control surfaces; δ n = [δ n1 δ n2 δ n3 δ n4 δ n5 δ n6 δ n7 δ n8 T represents the deflection angle of the control surfaces of the flying wing UAV, δ f = diag{δ f1 , δ f2 , …, δ f8} represents the stuck angle of the control surfaces; Γ n represents the nominal aerodynamic torque without control surface failure, Γ f represents the aerodynamic torque caused by the failure. In this embodiment, I x = 349.3, I y = 923.4, I z = 1321.7, I xz = 31.9. For the convenience of writing, the following definitions are introduced:

[0063]

[0064] where △ = I x I z - I xz 2 . Considering the above definitions and combining Equation (3), Equation (1) can be rewritten in the following form:

[0065]

[0066] where is the first derivative of Θ, is the first derivative of Ω, The expression is:

[0067] D A ​​= D + GΓ f ,

[0068] According to the attitude system dynamics (4), the second-order dynamics of the attitude angle can be obtained as follows:

[0069]

[0070] where is the first derivative of W, and D L is the lumped disturbance in the attitude tracking system, and its expression is:

[0071] D L = WD A ,

[0072] Define the attitude tracking error of the flying-wing UAV:

[0073]

[0074] where Θ d = [φ d θ d ψ d T , φ d , θ d , ψ d are the desired attitude angle commands respectively, and e φ , e θ , e ψ represent the roll, pitch, and yaw angle tracking errors respectively. Then, the tracking error dynamics of the attitude system can be obtained as follows:

[0075]

[0076] where and represent the second derivative and the first derivative of e Θ respectively, and and represent the second derivative and the first derivative of Θ d respectively.

[0077] Step 2: Design roll, pitch, and yaw three-channel observers based on the harmonic observer technology.

[0078] Design a harmonic observer for the disturbed attitude system (4) of the flying-wing UAV considering the actuator failure to estimate the lumped disturbance. The specific steps of the design include:

[0079] Design a three-channel harmonic observer for the disturbed attitude system (4) to estimate D A :

[0080] where, is the harmonic observer dynamics,, K​ω = diag{ω x , ω y , ω z} represents the frequency matrix of harmonic interference; are the derivatives of Z1, Z2, and Z3 respectively, is the estimated value of D A , and L1, L2, and L3 are the gains of the harmonic observer. Their specific forms are as follows:

[0081]

[0082] In this embodiment,

[0083] Step 3: Design a compound nonlinear dynamic inverse attitude controller for the flying wing UAV.

[0084] For the tracking error dynamics (6) of the flying wing UAV attitude system, combined with the interference estimation information of the harmonic observer construct a compound dynamic inverse controller. The specific steps of its design include:

[0085] Design a compound dynamic inverse controller for the tracking error dynamics (6) of the flying wing UAV attitude system:

[0086]

[0087] where is the estimated information of the multi-source interference D A , obtained from the harmonic observer (7), are the controller parameters, and their specific forms are as follows:

[0088]

[0089] where and are positive constants. In this embodiment,

[0090] Step 4: Based on the pseudo-inverse method, convert the desired torque into the aileron deflection of the flying wing UAV.

[0091] According to the virtual control quantity Γ n obtained in Step 3 and the aileron equation (3) of the flying wing UAV considering aileron failures, the actual control quantity δ n of the flying wing UAV attitude system is obtained by inverse solution based on the pseudo-inverse method:

[0092] δ n = B + Γ n ,

[0093] Among them, B + is the generalized inverse matrix of B. In this embodiment:

[0094]

[0095] In order to verify the anti-interference performance, command tracking performance, and fault tolerance performance of the method proposed in the present invention, considering multi-source interference and actuator failures, the control algorithm proposed in the present invention is simulated and verified based on the MATLAB simulation environment. During the simulation process, the initial values of the three attitude angles and angular velocities are respectively set as:

[0096] φ(0) = 0, θ(0) = 0, ψ(0) = 0, p(0) = 0, q(0) = 0, r(0) = 0;

[0097] In order to make the control task more challenging, the attitude angle commands are set in the following time-varying form:

[0098] φ d (t) = 3 + sin(t)°, θ d (t) = -2 + sin(t)°, ψ d (t) = 2 + sin(t)°;

[0099] Among them, t is time. During the simulation process, the external interference is set as:

[0100]

[0101] And the following fault scenarios are considered: there is no actuator failure from 0 to 10 seconds, the No. 2 actuator is damaged by 40% and the No. 7 actuator is damaged by 20% from 10 to 15 seconds, and the No. 5 actuator is stuck at 5° after 15 seconds.

[0102] The form and controller parameters of the composite nonlinear dynamic inverse controller based on harmonic observer (Harmonic Observer-based Composite Nonlinear Dynamic Inverse Controller, CNDIC+HO) of the method of the present invention have been given in the design example. The forms and controller parameter designs of the extended state observer (ESO) and the nonlinear dynamic inverse controller (NDCI) for comparison are as follows:

[0103] Design of the extended state observer:

[0104]

[0105] Among them and is the dynamic of the extended state observer, is the interference D L estimated value of, and are the derivatives of Z1 and Z2 respectively, and L1 and L2 are the observer gains, and their forms are as follows:

[0106]

[0107] where

[0108] Nonlinear dynamic inverse controller design:

[0109]

[0110] The form of its controller parameters is:

[0111]

[0112] where,

[0113] Figure 2 are the roll, pitch, and yaw three-channel attitude angle tracking error curves of the flying wing UAV under the action of the harmonic observer-based composite nonlinear dynamic inverse controller (CNDIC+HO) proposed by the present invention, the extended state observer-based composite nonlinear dynamic inverse control (CNDIC+ESO) for comparison, and the benchmark nonlinear dynamic inverse controller (NDIC), respectively; Figure 3 are the response curves of the control quantities of the roll, pitch, and yaw three channels of the flying wing UAV under the action of the CNDIC+HO, CNDIC+ESO, and NDCI proposed by the present invention. From Figure 2 、 Figure 3 it can be seen that the composite dynamic inverse control method based on the harmonic observer proposed by the present invention can achieve high-precision tracking of attitude commands under complex interference and fault conditions. The composite dynamic inverse control method based on the extended state observer for comparison only achieves high-precision tracking of attitude commands when the external interference is a constant value (0-5s); while the benchmark dynamic inverse control method cannot achieve high-precision tracking of attitude commands when there are interference and faults in the system. Figure 4 、 Figure 5They are the deflection angle response curves of the control surfaces 1-4 and 5-8 of the flying wing UAV, respectively. Figure 6 It is the estimation error response of the harmonic observer and the extended state observer to the disturbance. It can be seen that compared with the extended state observer, the harmonic observer can achieve high-precision estimation of the lumped disturbance.

[0114] The present invention realizes the asymptotic tracking of the reference commands of the roll, pitch, and yaw channel attitude angles of the flying wing UAV under the simultaneous action of periodic disturbances and actuator failures. Compared with the traditional dynamic inversion control method and the extended state observer technology, the present invention has better anti-interference ability and fault tolerance ability, and effectively suppresses the influence of harmonic interference and control surface failures on the control performance of the flying wing UAV.

[0115] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system, characterized in that, It includes the following steps: (1) Establish a perturbed dynamic model of the flying wing UAV attitude system affected by control surface faults and harmonic disturbances, and obtain the dynamic of the tracking error of the flying wing UAV attitude system; (2) Design roll, pitch, and yaw three-channel observers for the perturbed attitude system of the flying wing UAV based on harmonic observer technology; (3) For the dynamic of the tracking error of the flying wing UAV attitude system, combine the interference estimation information of the harmonic observer to construct a composite nonlinear dynamic inverse attitude controller for the flying wing UAV, and obtain the virtual control quantity; (4) Based on the pseudo-inverse method, convert the desired moment into the control surface deflection of the flying wing UAV; The implementation process of step (2) is as follows: Among them, Γ n represents the nominal aerodynamic moment under the condition of no control surface failure, is the harmonic observer dynamics, K ω = diag{ω x , ω y , ω z} represents the frequency matrix of harmonic interference; are the derivatives of Z1, Z2, and Z3 respectively, is the estimated value of D A , D A represents the lumped interference, and L1, L2, and L3 are the harmonic observer gains, and their specific forms are as follows:

2. A method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system according to claim 1, characterized in that The perturbed dynamic model of the flying wing UAV attitude system in step (1) is: where s φ , t θ , c φ and c θ represent sinφ, tanθ, cosφ and cosθ respectively; φ, θ and ψ represent the roll angle, pitch angle and yaw angle of the flying-wing UAV respectively, and represent the first-order derivatives of φ, θ and ψ respectively; p, q and r represent the rotational angular velocities of the flying-wing UAV about the body x, y and z axes respectively, and represent the first-order derivatives of p, q and r respectively; I x , I y and I z represent the moments of inertia of the flying-wing UAV about the body x, y and z axes respectively; I xz represents the product of inertia of the flying-wing UAV; τ x , τ y and τ z represent the aerodynamic moments of the flying-wing UAV about the body x, y and z axes respectively; D x , D y and D z represent the periodic disturbances acting on the body axes, and their dynamics satisfy: where ω x , ω y , ω z represent the frequencies of the three-axis disturbances, and their values are known; represents the internal dynamics of the disturbance, and represent respectively and the first-order derivatives of Through the moment in the perturbed dynamic model of the flying wing UAV attitude system, obtain the control surface loop model of the flying wing UAV including control surface faults: Γ = B[(I - K f )δ n + δ f = Bδ n + B(δ f - K f δ n ) = Γ n + Γ f (3) where Γ = [τ x τ y τ z T is the moment vector; denotes the nominal moment vector; denotes the moment vector caused by the influence of control surface failure; B ∈ R 3×8 denotes the control efficiency matrix of the control surface of the flying wing UAV; I ∈ R 8×8 is the identity matrix; K f = diag{k f1 , k f2 , …, k f8} represents the failure coefficient matrix of 8 control surfaces; δ n = [δ n1 δ n2 δ n3 δ n4 δ n5 δ n6 δ n7 δ n8 T represents the nominal value of the deflection angle of the control surface of the flying wing UAV ignoring the influence of failure, δ f = diag{δ f1 , δ f2 , …, δ f8} represents the stuck angle of the control surface; Γ n represents the nominal aerodynamic moment without control surface failure, Γ f represents the aerodynamic moment caused by the failure; The following definitions are introduced:​​ Where Δ=I x I z -I xz 2 Considering the above definitions, the simultaneous equations (3) and (1) can be rewritten as follows: wherein is the first derivative of Θ, is the first derivative of Ω, represents the lumped interference, and the expression is: D A = D + GΓ f According to the attitude system dynamics (4), the second-order dynamics of the attitude angle can be obtained: where is the first derivative of W, D L is the lumped interference including multi-source interference and the influence of control surface failure, and the expression is: D L = WD A Define the attitude tracking error of the flying wing UAV: where Θ d = [φ d θ d ψ d T , φ d , θ d , ψ d are the desired attitude angle commands respectively, e φ , e θ , e ψ represent the roll, pitch, and yaw angle tracking errors respectively. Then the attitude system tracking error dynamics are as follows:​ Among them and respectively represent the second derivative and the first derivative of e Θ , and and respectively represent the second derivative and the first derivative of Θ d .

3. A method for suppressing harmonic interference and active fault tolerance of the attitude system of a flying wing unmanned aerial vehicle according to claim 1, characterized in that, Step (3) is implemented through the following formula: Among them, is the estimated information of the lumped interference D A , and are controller parameters, and their specific forms are as follows: Among them, and are control parameters and are both positive constants.

4. A method for suppressing harmonic interference and active fault tolerance of a flying wing UAV attitude system according to claim 1, characterized in that, The implementation process of step (4) is as follows: The virtual control quantity Γ obtained according to step (3) n , according to the rudder loop equation, the true control quantity δ of the flying wing UAV attitude system is inversely solved based on the pseudo-inverse method n : δ n = B + Γ n Among them, B + is the generalized inverse matrix of B.

Images