Predictive control method and device for mode switching process of multi-mode turbine engine

Abstract

The present invention discloses a predictive control method for the mode switching process of a multi-mode turbine engine. This predictive control method divides the input variable geometry directional adjustment plan into two stages: main combustion chamber flameout and variable geometry adjustment mechanism directional adjustment. During the mode switching process, the multi-mode turbine engine is predictively controlled using a feedback correction loop consisting of an onboard adaptive composite model and a multi-mode predictive controller. The present invention also discloses a predictive control device for the mode switching process of a multi-mode turbine engine. Compared to existing technologies, the present invention can achieve precise control of thrust and flow during the mode switching process of a multi-mode turbine engine under conditions of engine performance degradation.

Description

Technical Field

[0001] The present invention belongs to the field of aerospace propulsion theory and engineering technology, and in particular relates to a predictive control method and device for a mode switching process of a multi-mode turbine engine. Background Art

[0002] The development of reusable aircraft with wide airspace and speed ranges has driven technological innovation in air-breathing propulsion. To improve aircraft range and economy, air-breathing propulsion systems must balance economic efficiency at low Mach numbers with high thrust performance at high Mach numbers. This has led to the emergence of a new generation of aircraft engines, represented by the XA-100 adaptive cycle engine, and hypersonic propulsion, represented by turboramjet combination engines. Both types of engines operate in turbofan mode at low flight speeds to reduce fuel consumption and improve economy, and in turbojet / ramjet mode at high flight speeds to maximize thrust performance at high Mach numbers. Having multiple operating modes is a key feature of the new generation of reusable aerospace propulsion systems, and controlling the mode switching process is a key challenge that must be addressed in the design of control systems for these engines.

[0003] The mode switching process is often accompanied by flow adjustment of variable geometry mechanisms such as mode selection valves and duct ejectors. This instantaneous switching of duct functions will cause sudden changes in the engine's thermal cycle, which is reflected in the engine performance output as large fluctuations in engine thrust and required flow. Since the mode switching process is often under high flight speed conditions, this instantaneous thrust and flow mutation can easily lead to aircraft instability. There may also be problems such as mode switching failure. Therefore, the study of mode switching process control methods is also a current research hotspot for variable cycle engines and turbine-based combined power.

[0004] Regarding the control methods for the mode switching process of variable cycle engines, Hao et al. proposed a method for optimizing adjustable parameters in the mode switching process with direct thrust control as the goal, and obtained the corresponding mode switching control law [A New Design Method for Mode Transition Control Law of Variable Cycle Engine[R]]. Zhang et al. proposed a general design method for the mode switching process based on particle swarm optimization and sensitivity calculation methods [General design method of control law for adaptive cycle engine mode transition[J]]. Zheng et al. proposed a design method for the "intermediate state" staged mode transition control law and an adjustment strategy for adjustable variables such as "mode transition speed" to achieve continuous engine thrust during the mode switching process [Research on the Design Method of the Mode Switch Transition Control Law of Adaptive Cycle Engine[J]]. Chen et al. proposed a constant flow mode switching method that ensures continuous engine flow during the mode switching process by controlling the fan static pressure [Flow control of double bypass variable cycle engine in modal transition[J]]. To achieve rapid mode switching, Wang et al. proposed a mode switching process control scheduling design method based on game optimization ideas [Game-Theory-Based Mode Switch Control Schedule Design for Variable Cycle Engine[J]]. Yu et al. established linear switching models under different modes and designed and planned corresponding smooth switching control strategies based on fuzzy control and intelligent algorithms [Active disturbance rejection control for uncertain nonlinear systems subject to magnitude and rate saturation: Application to aeroengine[J]].

[0005] Lv et al. have carried out extensive research on the control method of combined power mode switching process. A TBCC structural scheme based on liquid ammonia jet precooling [Thermodynamic modeling and analysis of ammonia injection pre-compressor cooling cycle: A novel scheme for high Mach number turbine engines[J]] and a variable-geometry ramjet combustor [Mode transition analysis of a turbine-based combined-cycle considering ammonia injection pre-compressor cooling and variable-geometry ram-combustor[J]] was proposed to improve the thrust performance of the turbofan engine during the mode switching process. A steady-state switching path that meets the requirements of constant thrust and lower fuel consumption during the mode switching process [Mode transition path optimization for turbine-based combined-cycle ramjet stage under uncertainty propagation of integrated airframe-propulsion system[J]] was obtained through a multi-objective optimization method. By introducing an improved neural network equilibrium manifold model and an expanded state observer, precise control of the liquid ammonia jet precooling temperature during the mode switching process [Intelligent ammonia precooling control for TBCC mode transition based on neural network improved equilibrium manifold expansion model[J]] was achieved. In order to ensure the safe and stable operation of the inlet and turbine engine during the mode switching process, a mode switching process control law design method considering the inlet deactivation margin and the combustion chamber outlet temperature is proposed [Mode transition control law analysis of ammonia MIPCCaeroengine considering inlet–compressor safety matching[J]].Xi et al. proposed a thrust augmentation method for the mode transition process based on an improved control plan [Design of thrust augmentation controls scheduled during mode transition for turbo-ramjet engine[J]]. Zheng et al. solved the thrust discontinuity problem during the mode transition process by optimizing the flight trajectory of the aircraft to change the thrust-drag characteristics [Trajectory optimization for a TBCC-powered supersonic vehicle with transition thrust pinch[J]].

[0006] The above research results provide rich theoretical support for the control of the mode switching process of multi-mode air-breathing propulsion systems. However, multi-mode air-breathing propulsion systems are used in reusable aircraft, and component performance degradation is inevitable during long-term service. Air-breathing propulsion systems are known to be strongly nonlinear systems. Degradation of engine component performance can lead to changes in the mapping relationship between engine control variables and thrust. Changes in the thermodynamic cycle characteristics during the mode switching process further exacerbate this nonlinear mapping uncertainty. For the aforementioned mode switching control law designed based on optimization methods, mode switching under performance degradation conditions cannot guarantee the desired continuous control targets of engine thrust and flow. This mode switching control method suffers from poor applicability and poor real-time performance. With current aircraft engine controller hardware, it is impossible to optimize the control variable commands within the control step. Therefore, it cannot be applied to real-time optimization control of the mode switching process, and mode switching can only be achieved through offline control laws. However, the control method based on an onboard adaptive model can directly use unmeasurable performance parameters such as engine thrust and flow as control targets, automatically compensating for the impact of engine performance degradation on control commands, and thus has promising application prospects in mode switching process control. However, since the mode switching process often does not only take place at one working point, there is a certain switching envelope, and the characteristics of multiple working modes and multiple adjustment variables increase the complexity and difficulty of airborne modeling of the mode switching process, which restricts the application of control methods based on airborne models in the field of mode switching process control. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to overcome the shortcomings of the existing technology and provide a predictive control method for the mode switching process of a multi-mode turbine engine, which can realize the precise control of thrust and flow during the mode switching process of a multi-mode turbine engine under the condition of engine performance degradation.

[0008] The technical solutions proposed in the present invention are as follows:

[0009] A predictive control method for a multi-mode turbine engine mode switching process divides an input variable geometry directional adjustment plan into two stages: main combustion chamber flameout and variable geometry adjustment mechanism directional adjustment; during the mode switching process, a feedback correction loop composed of an airborne adaptive composite model and a multi-mode predictive controller is used to predictively control the multi-mode turbine engine; the airborne adaptive composite model is used to perform real-time estimation of engine thrust and flow; the multi-mode predictive controller includes a turbofan mode predictive controller and a turbojet mode predictive controller, which are respectively used to predictively control the multi-mode turbine engine in the main combustion chamber flameout stage and the variable geometry adjustment mechanism directional adjustment stage; the turbofan mode predictive controller and the turbojet mode predictive controller both use the airborne adaptive composite model as a predictive model, perform online rolling optimization with the minimum engine thrust and flow change amplitude as the performance target, and generate engine control parameters during the mode switching process, thereby achieving a smooth transition of engine thrust and flow during the mode switching process.

[0010] Preferably, the airborne adaptive composite model includes a steady-state baseline model, a health parameter estimation module, and a propulsion system matrix module. The steady-state baseline model, the health parameter estimation module, and the propulsion system matrix module all include two states: a turbofan mode with the fuel flow in the main combustion chamber as the scheduling variable and a turbojet mode with the mode selection valve angle as the scheduling variable. The steady-state baseline model takes the flight altitude, Mach number, and engine adjustable variables in different working modes as input, and takes the engine steady-state performance parameters as output. It is obtained through training using a deep neural network method and is used to estimate the engine steady-state performance in real time. The health parameter estimation module is used to perform online real-time estimation of the engine performance degradation. The propulsion system matrix module stores a propulsion system matrix reflecting the influence of component performance degradation parameters on engine performance parameters corresponding to different modes and different baseline model working points established in advance offline, and is used to compensate and correct the engine steady-state performance parameters output by the steady-state baseline model according to the engine performance degradation estimated by the health parameter estimation module.

[0011] Further preferably, the health parameter estimation module is implemented by augmenting engine component performance degradation parameters into engine state space model state variables and designing a Kalman filter offline.

[0012] Preferably, the variation range of the engine control parameters used to establish the airborne adaptive composite model is obtained by pre-optimizing and solving the following optimization objective function offline:

[0013]

[0014] Where, F ,obj 、ma 2,objrepresent the engine thrust and flow before mode switching, respectively; F[k] and ma2[k] are the engine thrust and flow at the kth step during mode switching; u[k] is the control quantity at the kth step; ω1 and ω2 are weight coefficients.

[0015] Preferably, the online rolling optimization is achieved by optimizing and solving the following objective function:

[0016]

[0017] Where r is the engine control command, which is a constant value during the mode switching process; N p 、N u They are prediction time domain and control time domain respectively; is the estimated value of the command variable by the prediction model; Δu is the difference between the previous and next moments of the engine control quantity, that is, Δu(k+i)=u(k+i)-u(k+i-1); Q and R are semi-positive definite matrices.

[0018] Based on the same inventive idea, the following technical solutions can also be obtained:

[0019] A predictive control device for a multi-mode turbine engine mode switching process includes a feedback correction loop consisting of an onboard adaptive composite model and a multi-mode predictive controller, which is used to predictively control the multi-mode turbine engine during the mode switching process; the variable geometry directional adjustment plan input into the predictive control device is divided into two stages: main combustion chamber flameout and variable geometry adjustment mechanism directional adjustment; the onboard adaptive composite model is used to perform real-time estimation of engine thrust and flow; the multi-mode predictive controller includes a turbofan mode predictive controller and a turbojet mode predictive controller, which are respectively used to predictively control the multi-mode turbine engine in the main combustion chamber flameout stage and the variable geometry adjustment mechanism directional adjustment stage; the turbofan mode predictive controller and the turbojet mode predictive controller both use the onboard adaptive composite model as a predictive model, perform online rolling optimization with the minimum engine thrust and flow change amplitude as the performance target, and generate engine control parameters during the mode switching process, thereby achieving a smooth transition of engine thrust and flow during the mode switching process.

[0020] Preferably, the airborne adaptive composite model includes a steady-state baseline model, a health parameter estimation module, and a propulsion system matrix module. The steady-state baseline model, the health parameter estimation module, and the propulsion system matrix module all include two states: a turbofan mode with the fuel flow in the main combustion chamber as the scheduling variable and a turbojet mode with the mode selection valve angle as the scheduling variable. The steady-state baseline model takes the flight altitude, Mach number, and engine adjustable variables in different working modes as input, and takes the engine steady-state performance parameters as output. It is obtained through training using a deep neural network method and is used to estimate the engine steady-state performance in real time. The health parameter estimation module is used to perform online real-time estimation of the engine performance degradation. The propulsion system matrix module stores a propulsion system matrix reflecting the influence of component performance degradation parameters on engine performance parameters corresponding to different modes and different baseline model working points established in advance offline, and is used to compensate and correct the engine steady-state performance parameters output by the steady-state baseline model according to the engine performance degradation estimated by the health parameter estimation module.

[0021] Further preferably, the health parameter estimation module is implemented by augmenting engine component performance degradation parameters into engine state space model state variables and designing a Kalman filter offline.

[0022] Preferably, the variation range of the engine control parameters used to establish the airborne adaptive composite model is obtained by pre-optimizing and solving the following optimization objective function offline:

[0023]

[0024] Where, F ,obj 、ma 2,obj represent the engine thrust and flow before mode switching, respectively; F[k] and ma2[k] are the engine thrust and flow at the kth step during mode switching; u[k] is the control quantity at the kth step; ω1 and ω2 are weight coefficients.

[0025] Preferably, the online rolling optimization is achieved by optimizing and solving the following objective function:

[0026]

[0027] Where r is the engine control command, which is a constant value during the mode switching process; N p 、N u They are prediction time domain and control time domain respectively; is the estimated value of the command variable by the prediction model; Δu is the difference between the previous and next moments of the engine control quantity, that is, Δu(k+i)=u(k+i)-u(k+i-1); Q and R are semi-positive definite matrices.

[0028] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0029] (1) The present invention maintains accurate control of engine thrust and flow even when engine performance degrades, with the maximum thrust fluctuation being less than 0.66%, demonstrating wider applicability.

[0030] (2) The present invention not only has the same control effect as the traditional SQP optimization method, but also has an average single-step calculation time of only 2.8% of that of the SQP optimization method. It takes into account the thrust and flow control accuracy and real-time calculation of the mode switching process, and has better engineering application prospects in the field of multi-mode turbine engine mode switching control system design. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Design a process for the mode switching process predictive control method proposed by the present invention;

[0032] Figure 2 It is a multi-mode airborne adaptive composite model structure;

[0033] Figure 3 a predictive control structure for mode switching of a multi-mode turbine engine;

[0034] Figure 4(a) shows the baseline model engine thrust test error diagram (turbofan mode);

[0035] Figure 4(b) shows the error diagram of the baseline model engine flow test (turbofan mode);

[0036] Figure 5(a) shows the thrust accuracy comparison of the onboard model engine under the condition of component performance degradation (turbojet mode);

[0037] Figure 5(b) shows the flow accuracy comparison of the onboard model engine under the condition of component performance degradation (turbojet mode);

[0038] Figure 6(a) shows the estimation accuracy of the adaptive estimation module under the condition of single component performance degradation;

[0039] Figure 6(b) shows the estimation accuracy of the adaptive estimation module under the condition of multi-component performance degradation;

[0040] Figure 7(a) shows the engine thrust changes during the mode switching process;

[0041] Figure 7(b) shows the engine flow rate changes during the mode switching process;

[0042] Figure 8 Comparison of single-step time consumption of different control methods in the mode switching process;

[0043] Figure 9(a) shows the engine thrust changes during the mode switching process under the condition of component performance degradation;

[0044] Figure 9(b) shows the engine flow rate changes during the mode switching process when component performance degrades. DETAILED DESCRIPTION

[0045] In view of the shortcomings of the existing technology, the solution of the present invention is to divide the input variable geometry directional adjustment plan into two stages: main combustion chamber flameout and variable geometry adjustment mechanism directional adjustment. During the mode switching process, the feedback correction loop composed of the onboard adaptive composite model and the multi-mode predictive controller is used to perform predictive control on the multi-mode turbine engine, so as to achieve precise control of the thrust and flow during the mode switching process of the multi-mode turbine engine when the engine performance is degraded.

[0046] To facilitate public understanding, the following describes the technical solution of the present invention in detail by taking a certain type of turbine engine with a three-shaft, multi-combustion chamber configuration and turbojet / turbofan multi-working modes, which has a higher degree of nonlinearity, as an example, and combining it with the accompanying drawings:

[0047] The design process of the mode switching process predictive control method of this embodiment is as follows: Figure 1 As shown in the paper, firstly, the SQP optimization method is used to determine the variation range of the engine control parameters (main combustion chamber fuel flow, interstage combustion chamber fuel flow, afterburner fuel flow, nozzle throat area, etc.) that meet the requirement of unchanged engine thrust and flow during the mode switching process, thereby narrowing the modeling interval of the airborne model input parameters during the mode switching process; then, a modeling method of a multi-mode turbine engine airborne adaptive composite model is developed for the mode switching process. By establishing an airborne adaptive model of a multi-mode turbine engine, the engine thrust, flow, compression component stability margin and other performance parameters are output online, and the Kalman filter is used to estimate the engine performance degradation in real time. The output of the airborne model is compensated and corrected through the propulsion system matrix module; finally, the airborne adaptive composite model is used as a prediction model to design a predictive controller for the mode switching process, so as to achieve a smooth transition of the engine thrust and flow during the mode switching process.

[0048] In order to obtain the engine control parameter variation range for establishing an airborne adaptive composite model while ensuring constant engine thrust and flow, the present invention transforms the control parameter variation problem during the mode switching process into a single-objective, multi-constraint dynamic programming problem. The present invention determines the optimization objective function as shown in the following formula:

[0049]

[0050] Where, F ,obj 、ma 2,objwhere represents the engine thrust and flow rate before the mode switch, respectively. F[k] and ma2[k] represent the engine thrust and flow rate at step k during the mode switch. u[k] is the control variable at step k. ω1 and ω2 are weight coefficients. This objective function minimizes the changes in engine thrust and flow rate within each switching step.

[0051] The SQP method is used to solve the problem offline, so as to obtain the parameter variation range of the regulating variables such as the main combustion chamber, interstage combustion chamber, afterburner fuel flow, nozzle throat area, etc., which meets the conditions of unchanged engine thrust and flow during the mode switching process.

[0052] After determining the parameter variation range, a multi-mode turbine engine airborne adaptive composite model is further established. The multi-mode airborne adaptive composite model is the basis for predictive control of the mode switching process. Since the research object of the present invention has multiple modes and each mode has many adjustable variables (main combustion fuel flow, interstage combustion chamber fuel flow, afterburner fuel flow, nozzle throat area, etc.), the present invention proposes the following Figure 2 The multi-mode airborne adaptive composite model shown is mainly composed of three modules, namely the steady-state baseline model, the propulsion system matrix module, and the health parameter estimation module. Each module includes two states: turbofan mode and turbojet mode.

[0053] The steady-state baseline model uses flight altitude, Mach number, and engine adjustable variables in different operating modes as inputs, and outputs engine performance parameters. This model is trained and established using deep neural network methods to estimate engine steady-state performance in real time. The health parameter estimation module establishes state-space models for different modes offline, augments engine degradation parameters for each mode, and designs a Kalman filter to achieve online, real-time estimation of engine performance degradation and implement adaptive airborne model capabilities. The estimated engine performance degradation by the health parameter estimation module is transmitted to the propulsion system matrix. The propulsion system matrix then corrects the impact of engine component performance degradation on the steady-state baseline model output of the airborne model, achieving high-precision tracking and output of the airborne model performance parameters under engine performance degradation. The modeling process for each component is detailed below.

[0054] For the problem of multi-mode airborne composite model switching, the present invention establishes turbofan mode and turbojet mode airborne composite models within the mode switching interval respectively. In the main combustion chamber flameout stage during the mode switching process, the present invention selects the main combustion chamber fuel flow as the scheduling variable. When the main combustion chamber operates normally, the airborne composite model selects the turbofan mode model. When the main combustion chamber fuel flow is less than the main combustion chamber flameout boundary, it switches to the turbojet mode airborne composite model, and uses the mode selection valve angle as the scheduling variable to calculate the output value of the engine performance under different inputs.

[0055] This embodiment is based on the baseline modeling method of BP neural network. According to the working characteristics of different modes, the baseline models of turbofan mode and turbojet mode are established through offline training. The input parameters of the turbofan mode baseline model u=[H,Ma,m fb ,m fbin ,A8,m fa ] T The order is flight altitude, flight Mach number, main combustion chamber fuel flow, interstage combustion chamber fuel flow, nozzle throat area, afterburner fuel flow; output parameter y = [F,m a2 ,N f ,N CDFS ,N c ,S mf ,S mCDFS ,S mc ,T t48 ] T The following are engine thrust, engine flow, three rotor speeds, compression component surge margin, and interstage combustion chamber outlet total temperature. The input parameters of the turbojet model baseline model u=[H,Ma,m fbin ,A8,m fa ,α MSV ] T The flight altitude, flight Mach number, interstage combustion chamber fuel flow, nozzle throat area, afterburner fuel flow, mode selection valve angle are in turn; output parameter y = [F,m a2 ,S mf ,T t48 ,N f ] T They are engine thrust, engine flow, fan surge margin, total temperature at the interstage combustion chamber outlet, and low-pressure rotor speed.

[0056] Since the research object of this embodiment is a three-rotor structure with many rotating parts, in order to select the performance degradation parameters of the rotating parts, the present invention conducts a correlation analysis based on the literature and ultimately determines that the turbofan mode fan, core drive fan, compressor flow degradation, high, medium and low pressure turbine efficiency degradation, turbojet mode fan flow degradation, and low pressure turbine efficiency degradation are health estimation parameters. If these health parameters are considered when establishing the baseline model, there will be problems of excessive dimension and overly complex network. For this reason, based on the established baseline model, the present invention introduces a propulsion system matrix module to supplement the impact of engine performance degradation on the output parameters of the airborne model. The functional relationship between the control quantity and the output quantity can be expressed as:

[0057] ΔY=PΔu (2)

[0058] Where, turbofan mode ΔY=[ΔF,Δm a2 ,ΔN f ,ΔNCDFS ,ΔN c ,ΔS mf ,ΔS mCDFS ,ΔS mc ,ΔT t48 ] T (Turbojet mode ΔY=[ΔF,Δm a2 ,ΔN f ,ΔS mf ,ΔT t48 ] T ) is an m-dimensional vector, ΔY represents the deviation of the engine output parameter; Δu = [η F ,η CDFS ,η HC ,η HT ,η MT ,η LT ] T (Turbojet mode Δu=[η F ,η LT ] T ) is an n-dimensional vector, Δu represents the engine performance degradation parameter, and P is an m×n-dimensional propulsion system matrix element.

[0059] According to the performance degradation variables and output variables selected in this embodiment, the propulsion system matrix of the turbofan mode can be expressed as:

[0060]

[0061] The corresponding turbojet mode propulsion system matrix is:

[0062]

[0063] Through offline calculation, a propulsion system matrix of component performance degradation parameters corresponding to different modes and different baseline model working points is established to form a propulsion system matrix database, which is integrated into the airborne model to realize real-time compensation for performance degradation and achieve the purpose of airborne model adaptation.

[0064] To estimate the engine performance degradation under different modes, the present invention adopts the Kalman filter method. Its basic idea is to augment the performance degradation parameters of engine components into the state variables of the engine state space model, design a Kalman filter offline, establish a health parameter estimation module, integrate it into the airborne model, and use the deviation of the engine's measurable output parameters as the input of the filter. Then, the performance degradation of each engine component is estimated online, realizing real-time tracking of the airborne model in the actual state of the engine.

[0065] The mathematical expression of the engine state space model considering component performance degradation Δη and environmental noise is as follows:

[0066]

[0067] Where x represents the engine state. Considering that the state space model is also used in the prediction model of the model predictive control method, the turbofan mode x=[F,m a2 ,N f ,N CDFS ,N c ,S mf ,S mCDFS ,S mc ,T t48 ] T , control quantity u=[m fb ,m fbin ,A8,m fa ] T The output quantity composed of measurable parameters is y=[N f ,N CDFS ,N c ,T t6 ,P t3 ,P t6 ] T ; Turbojet mode x=[F,m a2 ,N f ,S mf ,T t48 ] T , control quantity u=[m fbin ,A8,m fa ] T The output quantity composed of measurable parameters is y=[N f ,T t6 ,P t6 ] T , A, B, C, D, L, M are dimension-appropriate matrices, and in turbofan mode, Δη=[Δη F ,Δη CDFS ,Δη HC ,Δη HT ,Δη MT ,Δη LT ] T (Turbojet mode Δη=[Δη F ,Δη LT ] T ), w and v are system noise and measurement noise respectively.

[0068] Since Δη cannot be obtained directly, here we take the performance degradation amount as the augmented state quantity based on formula (5), and the full-dimensional observer based on the Kalman filter can be obtained as follows:

[0069]

[0070] In the formula B k=[B 0] T ,C k =[CM], is an estimate that can be obtained from the airborne model, and the Kalman filter gain matrix K is determined by formula (7).

[0071]

[0072] Where P is the solution of the following Riccati equation, Q and R are the covariance matrices of the white noise matrices w and v, and I represents the identity matrix.

[0073] A k P+PA k T -PC k T R -1 C k P+Q=0 (8)

[0074] After the adaptive airborne composite model is established, the thrust and flow smooth transition control of the mode switching process based on the model predictive control method is studied. In view of the switching problem of the predictive model involved in the mode switching process, the present invention proposes Figure 3 The multi-mode predictive controller shown is used to implement mode switching process control.

[0075] The multi-mode turbine engine has two operating states: turbofan mode and turbojet mode. Under high Mach number operating conditions, the turbofan mode cannot meet the thrust requirements and needs to be switched to turbojet mode to further increase the engine thrust. The engine needs to shut down the main combustion chamber and adjust the variable geometry diverter ring and mode selection valve to achieve the switching of the turbofan to turbojet operating mode. Figure 3 The medium variable geometry directional control plan is divided into the main combustion chamber flameout (main combustion chamber fuel flow m fb The variable geometry adjustment mechanism directional adjustment refers to the mode selection valve (α MSV Adjustable from 0° to 90°) and variable geometry diverter ring (α splitter The multi-mode predictive controller in the figure includes a turbofan mode predictive controller and a turbojet mode predictive controller. The control objectives of the two controllers are the same, and both take the engine thrust and flow rate unchanged before and after the mode switching as the control objective (i.e. Figure 3Reference trajectory in multi-mode predictive controller). The present invention proposes to call different predictive controllers according to the variable geometry directional adjustment plan: in the main combustion chamber flameout stage, the turbofan mode predictive controller is called, and in the variable geometry adjustment mechanism adjustment process, the turbojet mode predictive controller is called. Since the engine thrust and flow are both unmeasurable parameters, the present invention establishes a multi-mode airborne adaptive composite model, and uses the airborne adaptive composite model to estimate the engine thrust and flow, so as to Figure 3 A feedback correction loop is formed in the predictive controller in the multi-mode airborne adaptive composite model. At the same time, the multi-mode airborne adaptive composite model is also the predictive model used by the rolling optimization module in the predictive control. The multi-mode predictive controller performs rolling optimization with the engine thrust and flow rate variation as the minimum performance target, and calculates the inter-stage combustion chamber fuel flow (m fbin ), afterburner fuel flow (m fa ), nozzle throat area (A8), and input them to the engine through the actuator, thereby realizing a complete predictive control loop for smooth transition of engine thrust and flow during mode switching.

[0076] like Figure 3 As shown in FIG, the control structure mainly consists of two core parts: a prediction model and an online rolling optimization module. The prediction model (a multi-mode airborne adaptive composite model) has been introduced in detail in the previous article. The online rolling optimization part is introduced below.

[0077] During the mode switching process, the desired control goal is to keep the engine thrust and flow rate continuous, and to ensure that the thrust and flow rate remain unchanged throughout the mode switching process as much as possible. Therefore, the present invention constructs the following objective function:

[0078]

[0079] Where r is the engine control command, which is a constant value during the mode switching process; N p is the prediction time domain, N u To control the time domain, is the estimated value of the command variable by the prediction model; Δu is the difference between the previous and next moments of the engine control quantity, that is, Δu(k+i)=u(k+i)-u(k+i-1); the first term of the objective function is to quickly track the command value, and the second term of the objective function is to ensure that the control quantity remains as stable as possible after the engine tracks the command; Q and R are semi-positive semi-definite matrices.

[0080] After determining the objective function, the online prediction optimization problem of the mode switching process is transformed into a quadratic programming problem, which can be solved using the active set method.

[0081] In order to verify the effect of the above-mentioned predictive control device, relevant simulation and analysis are carried out.

[0082] First, the accuracy of the airborne adaptive composite model established by the present invention is simulated and verified. According to the optimization of the mode switching process, for the turbofan mode, the modeling interval of the present invention is H = 20-24km, Ma = 3.75-4.2, and the main combustion chamber fuel flow m fb The range of change is 0.4~0kg / s, and the fuel flow rate of the interstage combustion chamber is m fbin The range of change is 1.2~1.8kg / s, the fuel flow rate of afterburner is m fa The range of change is 1.6~2.4kg / s, and the range of change of nozzle throat area A8 is 0.55~0.61m 2 For turbojet mode, the variable geometry splitter ring α MSV The variation range is 0 to 90°, and the variation ranges for the interstage combustor fuel flow, afterburner fuel flow, and nozzle throat area are the same as those for the turbofan mode. Figures 4(a) and 4(b) show that the baseline modeling accuracy of engine thrust and flow is high, with a maximum error of less than 1% and an average error of less than 0.035%.

[0083] To further verify the accuracy of the airborne adaptive composite model established in this invention, simulations of engine fan flow degradation and low-pressure turbine efficiency degradation were performed in turbojet mode. The performance degradation range was 0-3%. The accuracy comparison results of the airborne model output and engine component-level data are shown in Figures 5(a) and 5(b).

[0084] In Figures 5(a) and 5(b), the left column shows the accuracy of various parameters using an existing neural network-based airborne model under conditions of engine component performance degradation, while the right column shows the accuracy of various parameters using the airborne adaptive composite model proposed in this invention under conditions of engine component performance degradation. As can be seen from the figures, correcting engine performance degradation using the PSM matrix module effectively improves the accuracy of the airborne model and significantly enhances the modeling accuracy of engine thrust and flow. In turbojet mode, with engine component performance degradation, the calculation accuracy of each parameter of the airborne adaptive composite model exceeds 0.337%, providing a solid foundation for accurate estimation of engine performance parameters such as thrust and flow using model predictive control.

[0085] To verify the ability of the proposed airborne adaptive composite model to accurately track engine performance degradation, single-component performance degradation and simultaneous multi-component performance degradation simulations were conducted under the flight conditions of H = 24 km and Ma = 3.75. The results are shown in Figures 6(a) and 6(b). Figure 6(a) shows the simulation results of single-component performance degradation, where the fan flow rate is set to degrade by 2% at t = 0s and the low-pressure turbine efficiency is set to degrade by 2% at t = 40s. Figure 6(b) shows the simulation results of simultaneous gradual degradation of multi-component performance, where the fan flow rate, compressor flow rate, and intermediate-pressure turbine efficiency are set to degrade by 1% at t = 0s, and the core drive fan flow rate, high-pressure turbine efficiency, and low-pressure turbine efficiency are set to degrade by 1.5% at t = 40s. The simulation results show that the airborne adaptive composite model established in the present invention has good performance degradation tracking capability for both sudden performance degradation of a single component and gradual performance degradation of multiple components at the same time, has no steady-state estimation error, and shows good adaptive capability.

[0086] Figures 7(a) and 7(b) show the engine performance parameter responses during mode switching under flight conditions of H = 24 km, Ma = 3.75. In the legend, MPC represents the method proposed by the present invention, SQP represents the SQP optimization method, and PID represents the traditional PID control method. The engine thrust and flow responses in Figures 7(a) and 7(b) show that the SQP optimization method and MPC control method effectively ensure that the engine thrust and flow remain essentially unchanged during mode switching. The maximum thrust fluctuations for the PID control method, SQP optimization method, and MPC control method are 2.8%, 0.03%, and 0.73%, respectively. The maximum engine flow fluctuations for the three methods are 3.4%, 0.52%, and 0.31%, respectively.

[0087] Figure 8 A comparison of the computational time for a single step of different control methods is presented. The simulation was performed using an Intel Core i5-6500U 3.20GHz processor with 8GB of memory. While the SQP optimization method offers the best thrust control results, it also consumes the most computational time. Within a single control step of 25ms, real-time optimization is impossible with the computational processing power of current aircraft engine controllers. The MPC control method, on the other hand, has an average single-step computational time of 0.204ms. While this is nearly 50 times longer than the PID control method, it is significantly less than the SQP optimization method. Furthermore, the maximum single-step computational time is also relatively low, suggesting its potential for application in practical multi-mode turbine engine engineering control.

[0088] To further verify the applicability of the proposed predictive control method for the mode switching process, under the same flight conditions as the previous simulation, a 2% degradation of fan flow and a 2% degradation of low-pressure turbine efficiency were simultaneously set during the mode switching process. The control effects of the different control methods during the mode switching process under engine performance degradation are shown in Figures 9(a) and 9(b). As shown in Figures 9(a) and 9(b), the proposed MPC control method can still ensure that the engine thrust and flow remain essentially unchanged during the mode switching process under the condition of component performance degradation, with maximum changes of 0.66% and 1.71%, respectively. Under the condition of component performance degradation, the maximum thrust changes during the mode switching process for the PID control and SQP optimization methods are 2.0% and 0.28%, respectively, and the maximum engine flow changes are 2.8% and 1.71%, respectively. It is worth noting that for both the SQP optimization and MPC control methods, the maximum flow change occurs at the mode switching starting point, which is due to the set component performance degradation. For the entire switching process, both the PID control and SQP optimization methods have steady-state errors and cannot accurately track the control instructions during the mode switching process. However, the MPC control method has no steady-state errors, demonstrating that the mode switching predictive control method proposed in this invention has good adaptability to the performance degradation of engine components.

Claims

1. A predictive control method for a multi-mode turbine engine mode switching process, characterized in that: The input variable geometry directional regulation plan is divided into two stages: main combustion chamber flameout and variable geometry regulation mechanism directional regulation; in the process of mode switching, a feedback correction loop composed of an onboard adaptive composite model and a multi-mode predictive controller is used to predictively control the multi-mode turbine engine; the onboard adaptive composite model is used to perform real-time estimation of engine thrust and flow; the multi-mode predictive controller includes a turbofan mode predictive controller and a turbojet mode predictive controller for predictively controlling the multi-mode turbine engine in the main combustion chamber flameout stage and the variable geometry regulation mechanism directional regulation stage, respectively. The turbofan mode predictive controller and the turbojet mode predictive controller both use the onboard adaptive composite model as a prediction model and use the engine thrust and flow change amplitude as the prediction model. The onboard adaptive composite model includes a steady-state baseline model, a health parameter estimation module, and a propulsion system matrix module. The steady-state baseline model, the health parameter estimation module, and the propulsion system matrix module all include two states: a turbofan mode with the fuel flow in the main combustion chamber as the scheduling variable and a turbojet mode with the mode selection valve angle as the scheduling variable. The steady-state baseline model takes the flight altitude, Mach number, and engine adjustable variables in different working modes as inputs, and takes the engine steady-state performance parameters as outputs. The model is obtained through training using a deep neural network method and is used to estimate the steady-state performance of the engine in real time. The health parameter estimation module is used to perform online real-time estimation of engine performance degradation; the propulsion system matrix module stores a propulsion system matrix reflecting the influence of component performance degradation parameters on engine performance parameters corresponding to different modes and different baseline model operating points established offline in advance, and is used to compensate and correct the engine steady-state performance parameters output by the steady-state baseline model based on the engine performance degradation estimated by the health parameter estimation module.

2. The predictive control method for a multi-mode turbine engine mode switching process according to claim 1, characterized in that: The health parameter estimation module is implemented by augmenting the engine component performance degradation parameters into the engine state space model state variables and designing a Kalman filter offline.

3. The predictive control method for a multi-mode turbine engine mode switching process according to claim 1, wherein: The variation range of the engine control parameters used to establish the airborne adaptive composite model is obtained by pre-optimizing and solving the following optimization objective function offline: Where, F ,obj 、ma 2,obj represent the engine thrust and flow before mode switching, respectively; F[k] and ma2[k] are the engine thrust and flow at the kth step during mode switching; u[k] is the control quantity at the kth step; ω1 and ω2 are weight coefficients.

4. The predictive control method for a multi-mode turbine engine mode switching process according to claim 1, wherein: The online rolling optimization is achieved by optimizing and solving the following objective function: Where r is the engine control command, which is a constant value during the mode switching process; N p 、N u They are prediction time domain and control time domain respectively; is the estimated value of the instruction variable by the prediction model; Δu is the difference between the previous and next moments of the engine control quantity, that is, Δu(k+i)=u(k+i)-u(k+i-1); Q and R are semi-positive definite matrices.

5. A predictive control device for a multi-mode turbine engine mode switching process, characterized in that: The invention comprises a feedback correction loop composed of an onboard adaptive composite model and a multi-mode predictive controller, which is used for predicting and controlling a multi-mode turbine engine during mode switching; a variable geometry directional adjustment plan input into the predictive control device is divided into two stages: main combustion chamber flameout and variable geometry adjustment mechanism directional adjustment; the onboard adaptive composite model is used for real-time estimation of engine thrust and flow; the multi-mode predictive controller comprises a turbofan mode predictive controller and a turbojet mode predictive controller, which are respectively used for predicting and controlling the multi-mode turbine engine in the main combustion chamber flameout stage and the variable geometry adjustment mechanism directional adjustment stage; the turbofan mode predictive controller and the turbojet mode predictive controller both use the onboard adaptive composite model as a predictive model and the engine thrust as a predictive model. , and the flow rate change amplitude is minimized as the performance target to perform online rolling optimization, generate engine control parameters during the mode switching process, so as to achieve a smooth transition of engine thrust and flow during the mode switching process; the onboard adaptive composite model includes a steady-state baseline model, a health parameter estimation module, and a propulsion system matrix module, and the steady-state baseline model, the health parameter estimation module, and the propulsion system matrix module all include two states: a turbofan mode with the fuel flow rate of the main combustion chamber as the scheduling variable and a turbojet mode with the mode selection valve angle as the scheduling variable; the steady-state baseline model takes the flight altitude, Mach number, and engine adjustable variables under different working modes as input, and takes the engine steady-state performance parameters as output, and is obtained through training using a deep neural network method for real-time estimation of the engine steady-state performance; The health parameter estimation module is used to perform online real-time estimation of engine performance degradation; the propulsion system matrix module stores a propulsion system matrix reflecting the influence of component performance degradation parameters on engine performance parameters corresponding to different modes and different baseline model operating points established offline in advance, and is used to compensate and correct the engine steady-state performance parameters output by the steady-state baseline model based on the engine performance degradation estimated by the health parameter estimation module.

6. The predictive control device for a multi-mode turbine engine mode switching process according to claim 5, characterized in that: The health parameter estimation module is implemented by augmenting the engine component performance degradation parameters into the engine state space model state variables and designing a Kalman filter offline.

7. The predictive control device for a multi-mode turbine engine mode switching process according to claim 5, characterized in that: The variation range of the engine control parameters used to establish the airborne adaptive composite model is obtained by pre-optimizing and solving the following optimization objective function offline: Where, F ,obj 、ma 2,obj represent the engine thrust and flow before mode switching, respectively; F[k] and ma2[k] are the engine thrust and flow at the kth step during mode switching; u[k] is the control quantity at the kth step; ω1 and ω2 are weight coefficients.

8. The predictive control device for a multi-mode turbine engine mode switching process according to claim 5, characterized in that: The online rolling optimization is achieved by optimizing and solving the following objective function: Where r is the engine control command, which is a constant value during the mode switching process; N p 、N u They are prediction time domain and control time domain respectively; is the estimated value of the instruction variable by the prediction model; Δu is the difference between the previous and next moments of the engine control quantity, that is, Δu(k+i)=u(k+i)-u(k+i-1); Q and R are semi-positive definite matrices.

Images