A multivariable coordinated control method of intake port / engine for intake distortion

Abstract

The present invention discloses an intake duct / engine multivariable cooperative control method for intake distortion, including: deriving an iLQR algorithm based on a linear model, introducing rolling optimization on the basis of the algorithm to output only the first control quantity; using the control law, combined with the intake duct / engine coupling principle for intake distortion, designing an intake / engine multivariable closed-loop controller, and studying the performance of the system. The present invention applies the idea of ​​model predictive control to the common iLQR algorithm, taking the intake / engine state quantity and the control quantity as quadratic performance indicators, and augmenting the intake state quantity with the position of the common working point of the intake / engine, controlling the fuel flow, the critical cross-sectional area of ​​the tail nozzle, the second-stage inclined plate and the auxiliary exhaust valve angle, and designing an intake / engine integrated multivariable cooperative controller for intake distortion. The method uses online planning and solving in a time-based iterative calculation method to form a global dynamic optimization, solve the problem of model uncertainty, and improve the performance degradation caused by intake distortion.

Description

Technical Field

[0001] The invention belongs to the field of aero-engine simulation and control, and in particular relates to an intake duct / engine multivariable collaborative control method for intake distortion. Background Art

[0002] Intake flow field distortion often occurs when the aircraft is climbing or descending and facing various harsh and complex working environments. At this time, if the flow rate of the intake duct can match the engine, the engine can maintain normal operation. However, conventional intake duct control and engine control are independent of each other, and it is difficult to achieve efficient joint regulation of the intake duct and the engine. Therefore, in order to ensure that the intake duct and the engine are always well matched under intake distortion, it is necessary to consider the coupling effect between the intake duct and the engine on the basis of the intake duct / engine integrated mathematical model, and design a reliable intake duct / engine multivariable coordinated control system.

[0003] Linear Quadratic Regulator (LQR) refers to the optimization control problem of a dynamic system when the system under study is linear and the performance index is a quadratic function of the state variable and the control variable. The iLQR algorithm (iterative Linear Quadratic Regulator) is a model-based reinforcement learning algorithm that belongs to differential dynamic programming. The result obtained by calculation is the deviation of the optimal action from the current action, and multiple rounds of iterations are required during the update. This algorithm can effectively overcome problems such as model uncertainty and achieve excellent results in the control of the inverted pendulum.

[0004] When the aircraft is flying at high altitude and supersonic speed, the working state of the inlet is often in a deep subcritical state, and the inlet is prone to starting problems, which is particularly obvious under intake distortion. In this state, if the inlet flow is reduced by means of inlet bleed and other means to adjust the position of the end normal shock wave, the flow matching between the inlet and the engine can be greatly improved. Therefore, if the iLQR algorithm with excellent control tracking performance is adopted and the position of the inlet / engine common working point that can reflect the position of the end normal shock wave is incorporated into the controller design, the performance of the propulsion system can be improved.

[0005] However, the commonly used ordinary LQR control method has certain limitations. This control algorithm outputs the entire control trajectory at one time and acts entirely on the engine system. It lacks a dynamic control optimization process and does not consider re-collecting feedback information in real time. This has a certain impact on the control accuracy and robustness of the control system and requires certain improvements.

[0006] Therefore, it is necessary to design an efficient and stable inlet / engine multivariable iLQR controller to coordinate the inlet and engine for adjustment, so as to keep the propulsion system in a safe and reliable working state when the intake distortion interferes. Summary of the invention

[0007] The technical problem to be solved by the present invention is to address the defects of the background technology, propose a rolling optimization iLQR algorithm design control law based on a linear model, expand the common working point of the inlet duct / engine to the state quantity of the controller design, establish a turbofan inlet duct / engine high-pressure speed and pressure ratio multivariable controller for intake distortion, realize multivariable coordinated adjustment of the inlet / engine, solve the uncertainty problem of the model, and improve the performance degradation caused by intake distortion interference. Simulation shows that the controller has good steady-state and dynamic performance, and can prevent the inlet duct from entering an unstable state under intake distortion, and can significantly reduce the loss of engine thrust.

[0008] The present invention adopts the following technical solutions to solve the above technical problems:

[0009] An intake duct / engine multivariable coordinated control method for intake distortion comprises the following steps:

[0010] Step A), derive the iLQR algorithm based on the linear model, introduce rolling optimization, and only output the first control quantity.

[0011] Step B), using the derived control law, taking the engine state vector and the control vector as quadratic performance indicators, an intake duct / engine integrated multivariable collaborative controller for intake distortion is established, and the steady-state and dynamic performance of the controller is verified.

[0012] As a further optimization scheme of the intake duct / engine multivariable coordinated control method for intake distortion of the present invention, the specific steps of step A) are as follows:

[0013] Step A1), defining the loss function and state transfer function of the iLQR algorithm based on the linear model;

[0014] Step A2), reverse calculation iteratively obtains the controller gain parameters within the entire prediction period, and uses the solved controller gain parameters to forward calculate the optimal control sequence within the prediction period;

[0015] Step A3) introduces the idea of ​​rolling optimization in the model predictive control algorithm, uses the LPV model as a prediction model, and solves the optimal control sequence in a limited time domain starting from that moment, but only applies the control output at the current moment in the sequence to the system.

[0016] As a further optimization scheme of the intake duct / engine multivariable coordinated control method for intake distortion of the present invention, the specific steps of step B) are as follows:

[0017] Step B1), based on the intake duct / engine coupling mechanism for intake distortion, the control variables and the controlled variables are selected, and the position of the intake / engine common working point is augmented into the state quantity, and the loss function, state transfer function and control law of the multivariable iLQR controller based on the LPV model are derived using the engine augmented model with actuators;

[0018] Step B2), establishing a quadratic performance index based on the intake duct / engine state vector and the control vector, performing online rolling optimization on the control variable with the minimum of the index as the goal, and designing an intake duct / engine multivariable collaborative iLQR controller for intake distortion;

[0019] Step B3), after the engine / engine multivariable collaborative iLQR controller is designed, the performance of the operating point under intake distortion is verified with reference to the conventional engine / engine controller.

[0020] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0021] (1) The present invention proposes an intake duct / engine multivariable coordinated control method for intake distortion, which incorporates the position of the intake duct / engine common working point into the controller design, and can establish an intake duct / engine coupled multivariable closed-loop control system to achieve comprehensive regulation of the intake duct and the engine, so as to keep the propulsion system working safely and stably when the intake distortion interferes, thereby improving the performance of the engine;

[0022] (2) The inlet duct / engine multivariable collaborative controller proposed in the present invention can stably track any instruction, has a fast calculation speed, has good steady-state and dynamic performance, can overcome the uncertainty problem of the model, and improve the performance degradation caused by intake distortion interference, verifying the effectiveness of the control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a structural diagram of an intake duct / engine multivariable coordinated control method for intake distortion;

[0024] Figure 2 It is a design flow chart of an intake duct / engine multivariable collaborative controller for intake distortion;

[0025] Figure 3 This is a specific process diagram of the iLQR algorithm based on the linear model;

[0026] Figure 4 This is a specific process diagram of the rolling optimization iLQR algorithm based on the LPV model;

[0027] Figure 5 This is the change diagram of the intake angle of attack and total pressure distortion index under the disturbance of intake conditions at the altitude point H = 15km, Ma = 1.8;

[0028] Figure 6 This is the output speed tracking response diagram under the disturbance of intake conditions at the altitude point H = 15km, Ma = 1.8;

[0029] Figure 7 This is the output pressure ratio tracking response diagram under the disturbance of intake conditions at the altitude point H = 15km, Ma = 1.8;

[0030] Figure 8 This is the position change diagram of the common working point of the engine / engine under the disturbance of the intake conditions when the altitude is H = 15km and Ma = 1.8;

[0031] Fig. 9 This is the effective thrust variation diagram under the disturbance of intake conditions at the altitude point H = 15km, Ma = 1.8;

[0032] Fig.10 This is the change diagram of the intake angle of attack and total pressure distortion index under the disturbance of intake conditions at the altitude point H = 17km, Ma = 1.6;

[0033] Fig.11 This is the output speed tracking response diagram under the disturbance of intake conditions at the altitude point H = 17km, Ma = 1.6;

[0034] Fig.12 This is the output pressure ratio tracking response diagram under the disturbance of intake conditions at the altitude point H = 17km, Ma = 1.6;

[0035] Fig.13 This is the position change diagram of the common working point of the engine / engine under the disturbance of the intake conditions when the altitude is H = 17km and Ma = 1.6;

[0036] Fig.14 This is the effective thrust variation diagram when the air intake conditions disturb the altitude point H=17km and Ma=1.6. DETAILED DESCRIPTION

[0037] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0038] The idea of ​​the present invention is to first derive a rolling optimization iLQR algorithm based on a linear model, including defining a loss function and a state transfer function, iteratively calculating the controller gain parameters in the entire prediction period by reverse calculation, and using the solved controller parameters to forward calculate the optimal control sequence in the prediction period. Aiming at the requirements of multivariable coordinated control of the intake duct / engine under the interference of intake distortion, based on the integrated mathematical model of the intake duct / engine, the common working point of the intake duct / engine is expanded to the state quantity of the controller design, and a multivariable coordinated controller is established by combining the rolling optimization concept in the model predictive control with the derived iLQR multivariable control law. Compared with the traditional conventional intake / engine controller, the controller can effectively adjust the position of the common working point of the intake / engine through the second-stage inclined plate and the auxiliary exhaust valve, adjust the intake duct working state and the end positive shock wave, and improve the engine performance under intake distortion. Compared with other multivariable control methods, the control method has good tracking effect and fast calculation speed, overcomes the uncertainty problem of the model, and improves the performance degradation caused by environmental interference.

[0039] The specific implementation of the present invention takes the design of a multivariable coordinated iLQR controller for high pressure speed and pressure ratio of a certain type of turbofan engine as an example, and its control method structure is as follows: Figure 1 As shown, the design process is as follows Figure 2 As shown, the intake duct / engine multivariable coordinated control method includes the following steps:

[0040] Step A), derive the iLQR algorithm based on the linear model, introduce rolling optimization, and only output the first control quantity.

[0041] Step B), using the derived control law, taking the engine state vector and the control vector as quadratic performance indicators, an intake duct / engine integrated multivariable collaborative controller for intake distortion is established, and the steady-state and dynamic performance of the controller is verified.

[0042] The detailed steps of step A) are as follows:

[0043] Step A1), the state transfer function of the iLQR algorithm is:

[0044] x t+1 =f(x t ,u t )

[0045] In the formula, x t Represents the current state of the system, u t Represents the control amount at the current moment, x t+1 It represents the state of the system at the next moment.

[0046] The algorithm aims to minimize the long-term loss:

[0047]

[0048] Where c(x1,u1) represents the loss caused by executing the control amount u1 under the initial state x1. The control sequence {u1,u2,…u T The sum of the losses generated is Q, which represents the long-term loss of the entire cycle.

[0049] Therefore, the objective function can also be written as:

[0050]

[0051] The initial state x1 is known. When the state transition function is determined, the objective function is about the control sequence {u1,u2,…u T} function.

[0052] The control strategy π is defined as the mapping from state to control quantity. At time t, a state x t The state value function is defined as t =s when the expected value of long-term loss obtained by making control decisions according to the control strategy π:

[0053] V(x t )=E π [c(x t ,u t )+c(x t+1 ,u t+1 )+…+c(x T ,u T )|x t =s]

[0054] A state x at time t t The action value function is defined as t =s, take a certain control quantity u t =The expected value of long-term loss obtained by making control decisions based on the control strategy π after a:

[0055] Q(x t ,u t )=E π [c(x t ,u t )+c(x t+1 ,u t+2 )+…+c(x T ,u T )|x t =s,u t =a]

[0056] The iLQR algorithm based on the linear model has a loss function defined as a quadratic function and a state transfer function defined as a linear function:

[0057]

[0058]

[0059] In the formula, F=[F x F u ], f is the unknown coefficient matrix of the state transfer function, C xx , C xu , C ux , C uu are the quadratic coefficient matrices of the loss function, c x , c u are the coefficient matrices of the first-order terms of the loss function, F x , F u are the linear coefficient matrices of the state transfer function respectively.

[0060] The objective function is:

[0061]

[0062] Step A2), in order to find the control sequence {u1,u2,…u T}, solving partial derivatives is one of the feasible methods, but the amount of calculation is very large, and the partial derivatives of the control quantity at each moment are different. Therefore, it is considered to perform reverse iterative calculation to obtain the optimal control sequence in the entire prediction period. The specific process is as follows Figure 3 shown.

[0063] Step A21), first, if the prediction period is T, solve the action value function Q(x T ,u T ) Minimum control quantity u T :

[0064] At time t = T, the action value function Q(x T ,u T ) T Taking the derivative and setting it to 0 gives:

[0065]

[0066]

[0067] u T =-C xu -1 (C xu T x T +c u T )

[0068] Using the variable K t and k t To replace the expression of the control quantity, we get:

[0069]

[0070] So we have:

[0071] u T =K T x T +k T

[0072] Step A22), use x T Replace u T , get the value of x T The state value function V(x T ):

[0073] So x T The state value function is:

[0074]

[0075] Substitute the variables in the formula:

[0076]

[0077] get:

[0078]

[0079] Step A23), find the action value function Q(x) at time T-1 T-1 ,u T-1 ), and x T Replace with x T-1 and u T-1 Function: The action value function at time T-1 is:

[0080]

[0081] Where V(f(x T-1 ,u T-1 )) can be expanded to:

[0082]

[0083] Substituting the variables gives:

[0084] Q T-1 =C+F T V T F

[0085] qT-1 =c+F T V T f+F T v T

[0086] So we have:

[0087]

[0088] Then replace the time T with T-1, and repeat this step until the parameters of the entire cycle are solved.

[0089] Step A24), finally, the parameter K is solved t and k t The control sequence {u1,u2,…u T}:

[0090] u T =K T x T +k T .

[0091] In step A3), the concept of rolling optimization in the model predictive control method is introduced, the LPV model is used as the prediction model, and online rolling optimization is performed on the objective function. Only the first control quantity output in the control sequence within the solved prediction period is applied to the system to form a global dynamic optimization control effect.

[0092] The detailed steps of step B) are as follows:

[0093] Step B1), aircraft engines operate in a wide flight envelope, and the working environment often faces harsh and complex conditions, so the inlet outlet flow field is easily distorted. When the aircraft is climbing or descending, an angle of attack often occurs. As the angle of attack changes, the degree of distortion of the inlet outlet total pressure will also change accordingly. In the case of intake distortion, the working state of the inlet will deviate from the critical state. In order to ensure that the inlet and engine flow are always well matched, a reliable inlet / engine multivariable coordinated control system needs to be designed.

[0094] Based on the inlet / engine coupling mechanism, the controllable parameters of the inlet are studied. The state of the aircraft engine is x = [n H n L ] T , respectively high pressure speed and low pressure speed, control variable u=[W fb A8 δ2 γ] T , which are fuel flow, tail nozzle critical cross-sectional area, second stage ramp plate of inlet duct and auxiliary exhaust valve angle, output y = [n H EPR] T, respectively, are the high pressure speed and the pressure ratio, then the engine small deviation increment model is:

[0095]

[0096] Where A, B, C, D, E, and F are the state space matrices of the system, and ΔX is the position change of the forward / reverse common working point.

[0097] The actual position of the entry / exit common working point is:

[0098]

[0099] X=X0+ΔX

[0100] In the formula, and σ in Represent the throat flow coefficient and total pressure recovery coefficient of the inlet duct respectively, and σ0 represent the throat flow coefficient and total pressure recovery coefficient of the critical state engine / engine common working point respectively, and X0 is the position of the engine / engine common working point corresponding to the steady-state point.

[0101] The common working point of the inlet and the engine is a fictitious variable that can reflect the working state of the inlet. When the inlet / engine common working point is at different positions, the inlet is in different states. The common working point position X is less than 0, equal to 0 or greater than 0, indicating that the inlet is in a subcritical state, a critical state or a supercritical state. At the same time, in the case of subsonic speed, the common working point of the inlet and the engine only reflects the working state of the inlet; in the case of supersonic speed, the position of the common working point also reflects the position of the last positive shock wave in the inlet, which has its physical meaning. Because the inlet flow field may be distorted, the conventional inlet total pressure recovery coefficient is not selected as the controlled variable, and the inlet / engine common working point that can represent the working state of the inlet is selected as an important parameter for controller design.

[0102] Considering the characteristics of the fuel flow control device and the actuator, the state space equation of the actuator is:

[0103]

[0104] In the formula, A z , B z is the state space matrix of the actuator.

[0105] Augment the control variables into the engine model to form an augmented engine model with actuators:

[0106]

[0107] In the formula, x p =[Δn HΔn L ΔW fb ΔA8 Δδ2 Δγ] T , y p =Δy,X p =ΔX, C p =[CD],D p =0,E p =[EF],F p =0.

[0108] Differentiating this system yields:

[0109]

[0110] The error signal for the command is:

[0111]

[0112] In the formula, Represents the first-order differential of the command input.

[0113] The common working point location is:

[0114]

[0115] When the engine faces intake distortion, in order to ensure that the flow of the intake duct and the engine is matched and always works safely and reliably, it is hoped that the position of the intake / engine common working point will quickly return to the critical state, that is, X = 0, after deviating from the critical state. Therefore, it is considered to augment it into the state quantity of the controller design, and the state vector x p Further expand to get:

[0116]

[0117] Right now:

[0118]

[0119] In the formula,

[0120] Discretized into:

[0121]

[0122] In the formula, They are the continuous system state equations after discretization x t ,u t , The corresponding coefficient matrix is, Represents the first-order differential of the command input at the current moment.

[0123] The loss function is defined as a quadratic performance indicator:

[0124]

[0125] Where Q and R are the state vector weight coefficient matrix and control vector weight coefficient matrix in the quadratic performance indicator functional, respectively, and are taken as diagonal matrices.

[0126] The state transfer function is defined as:

[0127]

[0128] It is necessary to interpolate the current LPV model based on the scheduling parameters to obtain the state transfer function corresponding to the current state.

[0129] The control law is:

[0130]

[0131] In the formula, is the state quantity e, controller gain corresponding to X.

[0132] Therefore, we have:

[0133]

[0134] Step B2), combining the rolling optimization concept in the traditional iLQR algorithm and the model predictive control algorithm, taking the LPV model as the prediction model, establishing the quadratic performance index of the engine state vector and the control vector, repeating the online optimization with the goal of minimizing the index as much as possible, and finding the optimal control law in the finite control time domain from this moment on, and the optimization time domain continuously rolls forward with the advancement of time, thereby forming an online rolling optimization. The specific process is as follows Figure 4 shown.

[0135] Its performance indicators are:

[0136]

[0137] Establish a turbofan inlet / engine multivariable coordinated control structure such as Figure 1 As shown in the figure, the control structure first needs to interpolate the current LPV model according to the scheduling parameters to obtain the state transfer function corresponding to the current state, and iteratively solve the control sequence and controller parameters that minimize the performance index according to the reverse calculation in the prediction time domain, and only input the first control quantity into the intake duct / engine integrated object, and repeat the process at the next moment for re-planning, forming a rolling optimization.

[0138] Step B3), after the intake duct / engine multivariable collaborative iLQR controller is designed, the performance of the operating point under intake distortion is verified by comparing the intake duct and the engine with the conventional intake / engine controller that controls the intake duct and the engine independently.

[0139] The uniformity of the flow field at the inlet outlet is generally characterized by the inlet outlet distortion index, which is the ratio of the maximum difference in the total pressure at the inlet outlet to the average total pressure of the total outlet area.

[0140] When the aircraft is in the process of climbing or descending, the change of the inlet angle of attack often causes the change of the inlet outlet distortion index. Therefore, when the altitude point H = 15km and Ma = 1.8, the inlet angle of attack (AttackAngle) and the total pressure distortion index (DP) of the engine subjected to the disturbance of the inlet conditions change as follows: Figure 5 As shown in the figure, a high pressure speed and pressure ratio step command is given at 10s, an angle of attack and total pressure distortion index change is given at 20s, and it is restored at 35s. Affected by this kind of intake condition disturbance, the simulation results of tracking the high pressure speed and pressure ratio step command at the altitude point H = 15km, Ma = 1.8 are compared. Figure 6 , 7 , 8, and 9.

[0141] When the aircraft is operating in a wide flight envelope, it is likely to face a harsh and complex environment and be affected by intake distortion. Therefore, assuming that the altitude point H = 17km and Ma = 1.6, the intake angle of attack (AttackAngle) and the total pressure distortion index (DP) of the engine affected by the intake condition disturbance change as follows: Fig.10 As shown in the figure, at 10s, a high pressure speed and pressure ratio step command is given, at 20s, only the total pressure distortion index changes, and it is restored at 35s. Affected by this kind of intake condition disturbance, the simulation results of tracking the high pressure speed and pressure ratio step command at the altitude point H = 17km, Ma = 1.6 are compared. Fig.11 , 12 , 13, and 14.

[0142] As can be seen from the figure, under the two control schemes, the high-pressure speed and pressure ratio can stably track the command step signal at different altitudes without steady-state error. The dynamic performance comparison is shown in Table 1. According to the simulation results, compared with the conventional engine / engine control scheme, the engine / engine multivariable cooperative control scheme has a better dynamic response of tracking the high-pressure speed of the command step signal, is less affected by the intake condition disturbance signal, has a faster recovery time, and has a lower pressure ratio overshoot. The pressure ratio adjustment time is similar, and its recovery time from the intake condition disturbance signal is slightly longer. This is because the controller sacrifices dynamic performance to a certain extent in order to adjust the position of the common operating point of the engine / engine. Overall, the engine / engine multivariable cooperative controller has good steady-state and dynamic tracking performance.

[0143] Table 1 Comparison of dynamic performance of different high-altitude working point control schemes

[0144]

[0145] Since conventional intake / engine control does not consider the coupling relationship between intake duct control and engine control and the impact of intake distortion on the engine, the position of the intake / engine common working point deviates from the critical state, the intake duct and engine flow do not match, the intake duct total pressure recovery coefficient decreases, and the effective thrust of the engine decreases significantly. The designed intake / engine multivariable collaborative controller for intake distortion augments the algorithm state quantity of the intake / engine common working point position, and performs rolling optimization on the second-stage ramp and auxiliary exhaust valve angle in real time, such as Figure 8 , Fig.13 As shown in the figure, the common working point is effectively adjusted to a critical state, ensuring that the intake duct supply flow matches the engine. Although the recovery time of the pressure ratio due to intake disturbance increases slightly, the thrust loss caused by intake distortion is significantly reduced. Fig. 9 , Fig.14 As shown, under the influence of the corresponding intake distortion, when H=15km and Ma=1.8, the thrust loss is 13.12% of the original thrust level, while the conventional inlet / exit controller loses 24.16% of the original thrust level; when H=17km and Ma=1.6, the thrust loss is 12.26% of the original thrust level, while the conventional inlet / exit controller loses 16.58% of the original thrust level.

[0146] It can be seen that the inlet / engine multivariable collaborative controller can meet the requirements of inlet / engine multivariable control, has good control quality, can adapt to changes in various intake conditions of the engine, efficiently maintain the working state of the inlet, does not enter an unstable state under intake distortion, and ensures the reliable and safe operation of the propulsion system.

[0147] It should be pointed out that the above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes and substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An intake duct / engine multivariable coordinated control method for intake distortion, characterized in that: The following steps are involved: Step A), derive the iLQR algorithm based on the linear model, introduce rolling optimization, and only output the first control quantity; Step B), using the derived control law, taking the engine state vector and the control vector as quadratic performance indicators, establishing an intake duct / engine integrated multivariable collaborative controller for intake distortion, and verifying the steady-state and dynamic performance of the controller; The specific steps of step A) are as follows: Step A1), defining the loss function and state transfer function of the iLQR algorithm based on the linear model; Step A2), reverse calculation iteratively obtains the controller gain parameters within the entire prediction period, and uses the solved controller gain parameters to forward calculate the optimal control sequence within the prediction period; Step A3), introduce the idea of ​​rolling optimization in the model predictive control algorithm, use the LPV model as the prediction model, solve the optimal control sequence in a limited time domain from the current moment, but only apply the control output at the current moment in the sequence to the system; The specific steps of step B) are as follows: Step B1), based on the intake duct / engine coupling mechanism for intake distortion, the control variables and the controlled variables are selected, and the position of the intake / engine common working point is augmented into the state quantity, and the loss function, state transfer function and control law of the multivariable iLQR controller based on the LPV model are derived using the engine augmented model with actuators; Step B2), establishing a quadratic performance index based on the intake duct / engine state vector and the control vector, performing online rolling optimization on the control variable with the minimum of the index as the goal, and designing an intake duct / engine multivariable collaborative iLQR controller for intake distortion; Step B3), after the engine / engine multivariable collaborative iLQR controller is designed, the performance of the operating point under intake distortion is verified with reference to the conventional engine / engine controller.

2. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: The loss function and state transfer function of the iLQR algorithm based on the linear model in step A1) are: In the formula, x t Represents the current state of the system, u t represents the control amount at the current moment, x t+1 It represents the state of the system at the next moment. F=[F x F u ], f is the unknown coefficient matrix of the state transfer function, C xx , C xu , C ux , C uu are the quadratic coefficient matrices of the loss function, c x , c u are the coefficient matrices of the first-order terms of the loss function, F x , F u are the linear coefficient matrices of the state transfer function respectively; The objective function is: Where c(x1,u1) represents the loss caused by executing the control amount u1 under the initial state x1. The control sequence {u1,u2,…u T The sum of the losses generated is Q, which represents the long-term loss of the entire cycle.

3. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 2, characterized in that: Step A2) comprises: Step A21), first, if the prediction period is T, solve the action value function Q(x T ,u T ) Minimum control quantity u T : At time t = T, the action value function Q(x T ,u T ) T Taking the derivative and setting it to 0 gives: u T =-C xu -1 (C xu T x T +c u T ) Using the variable K t and k t To replace the expression of the control quantity, we get: So we have: u T =K T x T +k T Step A22), use x T Replace u T , get the value of x T The state value function V(x T ): So x T The state value function is: Substitute the variables in the formula: get: Step A23), find the action value function Q(x) at time T-1 T-1 ,u T-1 ), and x T Replace with x T-1 and u T-1 Function: The action value function at time T-1 is: Where V(f(x T-1 ,u T-1 )) can be expanded to: Substituting the variables gives: Q T-1 =C+F T V T F q T-1 =c+F T V T f+F T v T So we have: Then replace time T with T-1, and repeat this step until the parameters of the entire cycle are solved; Step A24), finally, the parameter K is solved t and k t The control sequence {u1,u2,…u T }: u T =K T x T +k T 。 4. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: The control variables in step B1) include: fuel flow W fb , tail nozzle critical cross-sectional area A8, the second stage inclined plate angle δ2 and the auxiliary exhaust valve angle γ of the inlet; the controlled variables, i.e. the output, include: high pressure speed n H And pressure ratio EPR.

5. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: The actual position of the forward / reverse common working point augmented to the state quantity designed by the controller in step B1) is: X=X0+ΔX In the formula, and σ in Represent the throat flow coefficient and total pressure recovery coefficient of the inlet duct respectively, and σ0 represent the throat flow coefficient and total pressure recovery coefficient of the critical intake / engine common working point respectively, X0 is the intake / engine common working point position corresponding to the steady-state point; when the intake / engine common working point position X is less than 0, equal to 0 or greater than 0, it means that the intake duct is in a subcritical state, a critical state or a supercritical state respectively.

6. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: The loss function established in step B1) is: Where Q and R are the state vector weight coefficient matrix and control vector weight coefficient matrix in the quadratic performance index functional, respectively, and are taken as diagonal matrices; The state transfer function is: In the formula, They are the continuous system state equations after discretization x t ,u t , The corresponding coefficient matrix is, Represents the first-order differential of the command input at the current moment; The control law is: In the formula, The state vector x, x designed for the forward / backward multivariable cooperative controller is p =[Δn H Δn L ΔW fb ΔA8 Δδ2 Δγ] T , n L is the low pressure speed, e=ry, r represents the command input, y=[n H EPR] T , X is the actual position of the forward / forward common working point, is the state quantity e, controller gain corresponding to X.

7. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: The quadratic performance index established in step B2) is: In the formula, the state vector of the design of the forward / forward multivariable cooperative controller is Control vector 8. The intake duct / engine multivariable coordinated control method for intake distortion according to claim 1, characterized in that: After obtaining the controlled quantity instruction at each moment, the multivariable collaborative iLQR controller reversely calculates and iterates to obtain the controller gain in the entire prediction period, and uses the solved controller parameters to forward calculate the optimal control sequence in the prediction period, but only inputs the first control quantity into the inlet duct / engine integrated object, and performs online rolling optimization on the control quantity in combination with the quadratic performance index of the state quantity and the control quantity, and augments the position of the common working point of the inlet duct / engine into the state quantity to realize the inlet duct / engine multivariable collaborative control, which can ensure the safe and reliable operation of the propulsion system under intake distortion.

Images