Model-free smooth transition control method for aero-engine mode switching

Abstract

The invention provides a model-free smooth transition control method for aero-engine mode switching, and belongs to the field of aero-engine control. Firstly, a design method based on linear active disturbance rejection control is introduced into an aero-engine control system and used for estimating and compensating unmodeled dynamic and external disturbance in the aero-engine control system in real time. And dividing the time scheduling linear active-disturbance-rejection controller into a transition controller and a steady-state controller according to the operation time interval. And by considering the actual operation time interval of the aero-engine control system, time scheduling parameters are introduced to dynamically adjust the maximum update times of the operation time interval of the time scheduling linear active disturbance rejection controller. And finally, for the switching dynamic error system of time scheduling, deducing a sufficient condition that the switching dynamic error system of time scheduling can keep asymptotically stable under the constraint of average residence time by performing stability analysis and designing that time and switching signals depend on a multi-discontinuous Lyapunov function.

Description

Technical Field

[0001] This invention belongs to the field of aero-engine control and relates to a model-free smooth transition control method for aero-engine mode switching. Background Technology

[0002] As the core power plant of an aircraft, the performance of an aero-engine directly affects the overall flight performance and safety of the aircraft. An aero-engine is a highly complex aero-thermodynamic coupling system, typically composed of multiple components such as the air intake, fan, compressor, combustion chamber, turbine, and exhaust nozzle. During operation, it exhibits strong nonlinearity, strong coupling, and multi-modal operating characteristics. In high-speed aircraft, the stability of aero-engine thrust output is of paramount importance. During the switching of aero-engine operating modes, significant fluctuations in fuel quantity and flow rate occur at the moment of transition, affecting thrust stability and potentially leading to engine failure, thus compromising the safety and reliability of the aircraft.

[0003] With its wide adjustable parameter range and strong adaptability, the active disturbance rejection controller (ADRC) can improve the stability of aero-engine control and is a classic control method in the field of tracking control. During the transition between operating modes of an aero-engine, controller switching can trigger significant turbulence in the transition state, affecting the engine's stability and reliability. However, due to the more complex switching feedback control law of the switching ADRC compared to other feedback control methods, general disturbance-free switching methods cannot effectively solve the problem of disturbance-free switching in linear ADRC. We propose a linear ADRC disturbance-free switching controller, which divides the controller into a transition controller and a steady-state controller. A weighted function transition mechanism is introduced between adjacent mode control laws, allowing the control input and system output to evolve continuously at the moment of switching, avoiding abrupt changes. However, aero-engines may be affected by unknown factors such as external disturbances during mode switching, resulting in variable actual operating times for each mode. This leads to a new consideration: can the influence of the actual switching signal be taken into account, achieving coordinated design of the switching signal and the operating time of the transition controller? This would allow for dynamic adjustment of the transition controller's operating time based on the actual system operating time, providing greater flexibility for the design and practical application of the linear ADRC disturbance-free switching controller. Summary of the Invention

[0004] To address the aforementioned issues, this invention employs a time-scheduled linear active disturbance rejection (ADNR) and disturbance-free switching strategy. The transient controller is segmented over time, resetting to zero when the system's operating mode changes, and then gradually increasing in a stepwise manner to the maximum allowable value within the same operating mode. This strategy, through the coordinated design of the transient controller's runtime and switching signals, not only effectively overcomes the strict limitations on transient controller runtime found in previous studies but also significantly expands its applicability, improves control performance, and reduces conservatism.

[0005] The technical solution of this invention: A model-free smooth transition control method for aero-engine mode switching, comprising the following steps: Step 1: Modeling the aircraft engine switching system: A design method based on linear active disturbance rejection control is introduced into the aero-engine control system to estimate and compensate for unmodeled dynamic and external disturbances in the aero-engine control system in real time.

[0006] Step 2: Construct a time-scheduled linear disturbance-free switching active disturbance rejection controller: The time-scheduled linear disturbance rejection controller (RTDC) is abbreviated as time-scheduled linear RDC. Based on the operating time interval, the time-scheduled RDC is divided into a transient controller and a steady-state controller. Furthermore, by considering the actual operating time interval of the aero-engine control system, a time-scheduling parameter is introduced to dynamically adjust the maximum number of updates within the operating time interval of the time-scheduled RDC.

[0007] Step 3: Stability Analysis: By conducting stability analysis on the switching dynamic error system of time scheduling and designing time- and switching signal-dependent discontinuous Lyapunov functions, sufficient conditions are derived for the switching dynamic error system of time scheduling to maintain asymptotic stability under the average dwell time constraint.

[0008] The beneficial effects of this invention are: This invention addresses the problem of significant fuel fluctuations during mode switching in aero-engines, which negatively impact aircraft safety and reliability. It proposes a model-free smooth transition control method for aero-engine mode switching. This method utilizes a controller that dynamically adjusts based on the actual operating time of the aero-engine mode, enabling it to better adapt to external disturbances and operating condition changes encountered by the aero-engine in complex flight environments. The method has proven effective in aero-engine control, reducing input value jumps during system switching, effectively suppressing fuel fluctuations and thrust ripples, achieving a smooth transition in system output, and mitigating the adverse effects of sudden input changes. In conclusion, this invention provides a solid theoretical basis and feasible engineering support for the safe and reliable operation of aero-engines. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the subsystems of an aero-engine control system; Figure 2 It is a switching signal; Figure 3 These are the high-voltage rotor speed state response curves under different methods; Figure 4 This is a comparison chart of fuel consumption under different methods. Detailed Implementation

[0010] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0011] A model-free smooth transition control method for aero-engine mode switching, comprising the following steps: Step 1: Modeling the aircraft engine switching system: Due to the strong coupling and nonlinearity among the internal components of aero-engine control systems, obtaining accurate mathematical models is often difficult, posing challenges to controller design. This invention introduces a design method based on linear active disturbance rejection control. This method does not rely on an accurate mathematical model of the controlled object and can estimate and compensate for unmodeled dynamics and external disturbances in the aero-engine control system in real time. The mathematical expression of the aero-engine control system is described below: in, These are the state variables of an aero-engine, including the low-pressure rotor speed. High-voltage rotor speed Fuel flow . and These are the dynamic characteristics of the aero-engine control system and the nonlinear functions of external disturbances, respectively. It is the measurement output of the aircraft engine control system. The coefficients are known. yes The first derivative.

[0012] Within the flight envelope of an aero-engine, its operating state changes continuously over time and is divided into multiple subsystems. Indicates the time of the aircraft engine control system The operating mode. ,in It is the number of subsystems. , Represents positive integers. Let represent the set of nonnegative real numbers.

[0013] Linear active disturbance rejection control (ADRC) treats the coupling effects between channels and external disturbances caused by the environment as the total disturbance of the aero-engine control system. Each channel independently estimates the total disturbance in real time through time-scheduled switching of the extended state observer. Linear ADRC exhibits good control performance in aero-engine control systems. The total disturbance is defined as... The mathematical expression for the aero-engine control system is rewritten as follows: in, , , , express The derivative of, and let The amplitude is finite, that is, there exists a... Makes all have .

[0014] Step 2: Construct a time-scheduled linear disturbance-free switching active disturbance rejection controller: The time-scheduled linear active disturbance rejection controller (ADRC) is referred to as the time-scheduled linear ADRC. During the operating time interval... The time-scheduled linear active disturbance rejection controller is divided into a transient controller and a steady-state controller. The operating time interval... and The internal controller is a transition controller, while the running time interval... The internal controller is a steady-state controller. This represents the segmentation score of the transition controller in the i-th subsystem. This represents a pre-defined fixed period. Furthermore, by considering the actual operating time interval of the aero-engine control system, a time scheduling parameter is introduced. To dynamically adjust the maximum number of updates within the operating time interval of the time-scheduled linear active disturbance rejection controller. Represents a positive integer. The time scheduling parameters of the time-scheduled linear active disturbance rejection controller are calculated using the following equation: The transient controller and steady-state controller of the time-scheduled linear active disturbance rejection controller both consist of two parts: a time-scheduled switching extended state observer and a time-scheduled switching feedback control law.

[0015] (2.1) Time-scheduled switching extended state observer: in, and respectively and The estimated value, Indicates time The high-pressure rotor speed of the aircraft engine, Indicates time The total disturbance. and The time-scheduled switching extended state observer's first and second order gain parameters are adjusted according to the following rules: ,and It is the bandwidth of the linear-time scheduling switching extended state observer. This represents the control input gain coefficient. Indicates control input; (2.2) Time scheduling switching feedback control law: in, It is the reference input signal. It is the gain of the active disturbance rejection controller.

[0016] In practical aero-engine control systems, switching signals exhibit dynamic uncertainty. During mode transitions, unforeseen factors such as changes in flight missions may occur, leading to inconsistencies in the actual operating times of different modes. To address this issue, the time-scheduled linear disturbance-free switching active disturbance rejection controller described above not only maintains the continuity of control gain but also considers the impact of the actual switching signal. The operating time interval of the transition controller is dynamically adjusted based on the actual operating time of the aero-engine control system. Regarding the operating time interval... The signal for switching operating modes is and ,in Proposed for use from subsystem Switch to subsystem The time-scheduled variable continuous gain switching controller has the following formula: in, Is The time scheduling parameters at any given moment For the subsystem before switching At the switching time The controller gain; The time-scheduling switching feedback control law is rewritten as follows: To facilitate the analysis of the stability of the control system, let , and These represent tracking error, state estimation error, and total disturbance estimation error, respectively.

[0017] in, The derivative of r(t); Represents time t The derivative; Therefore, the dynamic error system for time scheduling switching is represented as: Where the state vector , , It is a column vector; Step 3: Stability Analysis: By conducting stability analysis on the switching dynamic error system of time scheduling and designing time- and switching signal-dependent discontinuous Lyapunov functions, sufficient conditions are derived for the switching dynamic error system of time scheduling to maintain asymptotic stability under the average dwell time constraint.

[0018] First, construct the time- and switching signal-dependent multidiscontinuous Lyapunov functions:

[0019] Based on this, define a matrix: in , This indicates that the diagonal matrix has parameters. ; and yes elements, It is a piecewise linear interpolation function, specifically in the form of: .

[0020] (3.1) The running time interval is hour, This represents the decay rate of the Lyapunov function of the transient controller; in, The derivatives of the constructed time- and switching signal-dependent discontinuous Lyapunov functions; This indicates that under subsystem i, the scheduling parameters are based on the current time. The system matrix in the switching dynamic error system determined by the selected values; From the above equation, we get: in, Represents symmetrical elements. , Further results were obtained: in, , .

[0021] according to ,get: Similarly: If the following conditions are met: Then it satisfies .

[0022] (3.2) The running time interval is hour: Its proof method and running time range are as follows: The times are the same, therefore the following condition is satisfied: Then it is guaranteed that: (3.3) The running time interval is hour, The decay rate of the multidiscontinuous Lyapunov function representing the transient controller is expressed by: To satisfy ; On the other hand, consider the subsystem j At any moment Switch to subsystem i and subsystem j During the running time interval Internal execution. Definition Running time interval The length of. Then, if ,inequality , Indicates the first j Each subsystem transition controller is divided into fractions. It is a constant that satisfies This is used to describe the jump in the Lyapunov function value at the switching moment. Indicates rounding down.

[0023] This means: Combination By definition, we obtain: in, For the first The left limit of each switching moment, that is, the instant before the switching occurs; definition Running time range The transient controller runtime is set as follows: Analyze the overall non-incremental characteristics.

[0024] in, Indicates the current time The Dolyapnov function of the activated subsystem, To indicate the system at the switching moment The state value, (·) is an exponential function. It is in the interval The total number of handovers; n is an integer variable used only to accumulate the time difference between adjacent handover moments.

[0025] Scaling the above formula: in, , To represent the system from the initial time... up to the current moment Total runtime; This refers to the average length of stay. For jitter boundaries.

[0026] For a switching signal and If a positive number exists ,and This makes the following equation true: Then, when the average length of stay meets the following conditions:

[0027] hour, Converging to zero, the dynamic error system of time scheduling switching is affected by disturbances. The situation is gradually stabilizing.

[0028] Step 4: Stability analysis of the smooth transition experiment of the aero-engine: In the control process of aero-engine control systems, frequent changes in operating conditions and the significant time-varying nonlinear characteristics of the aero-engine control system itself can easily lead to problems such as severe signal fluctuations, which can even affect the stable operation of the aero-engine control system in severe cases. Typically, the entire flight envelope is divided into multiple characteristic operating condition intervals, and a corresponding controller is designed for each interval to meet the dynamic performance requirements of the aircraft at different mission stages. When the flight mission changes and the target trajectory needs to cross multiple flight regions with significant nonlinear differences, the controller needs to dynamically switch to meet the changes in the characteristics of the aero-engine control system. We selected three nonlinear intervals of the aero-engine and designed a time-scheduled linear disturbance-free switching active disturbance rejection controller to verify the effectiveness of the proposed smoothing strategy. The parameter selection for the time-scheduled linear disturbance-free switching active disturbance rejection controller is as follows: The proposed method: Subsystem 1: , , , , , , , ; Subsystem 2: , , , , , , ; Subsystem 3: , , , , , , , ; There is no disturbless handover strategy: , , , , , ; Based on the parameters mentioned above, we obtain =1.10. It should be noted that the runtime interval of the transition controller and the number of segments within that interval directly affect the smoothing effect.

[0029] During the running time interval Within the subsystem, subsystem 2 first operates the transition controller, whose gain is updated sequentially according to the time-scheduled continuous gain switching controller; that is, the gain is first used... , and then updated to When the switching signal is generated, subsystem 2 directly switches from the transition controller to the transition controller of subsystem 3. Afterwards, subsystem 3 operates for 1 second, during which its controller gain is sequentially switched according to a time-scheduled continuous gain switching mechanism. , and steady-state controller gain The order of updates.

[0030] As can be seen from the above process, the update of the controller gain of the aero-engine control system is related to the actual switching signal of the aero-engine control system. The running time of the transition controller can be dynamically adjusted according to the actual running time of the aero-engine control system. This method effectively overcomes the limitation on the running time of the transition controller and greatly expands the applicable scenarios while reducing its conservatism.

[0031] In summary, the proposed strategy can achieve stable fuel supply to the aero-engine control system under multiple operating conditions. As shown in the figure below, for different methods, the proposed method effectively suppresses sudden changes in fuel quantity at the switching point in the fuel flow comparison. Meanwhile, the different methods... The response curves further demonstrate that the system's state trajectory is smoother at the switching moments. The method proposed in this invention can more flexibly respond to random switching signals and achieve a smooth transition in the system output.

Claims

1. A model-free smooth transition control method for aero-engine mode switching, characterized in that, The steps are as follows: Step 1: Modeling the aircraft engine switching system: A design method based on linear active disturbance rejection control is introduced into the aero-engine control system for real-time estimation and compensation of unmodeled dynamic and external disturbances in the aero-engine control system. Step 2: Construct a time-scheduled linear disturbance-free switching active disturbance rejection controller: The time-scheduled linear disturbance-free switching active disturbance rejection controller (ADRC) is abbreviated as time-scheduled linear ADRC. Based on the operating time interval, the time-scheduled linear ADRC is divided into a transient controller and a steady-state controller. By considering the actual operating time interval of the aero-engine control system, time-scheduling parameters are introduced to dynamically adjust the maximum number of updates of the operating time interval of the time-scheduled linear ADRC. Step 3: Stability Analysis: By conducting stability analysis on the switching dynamic error system of time scheduling and designing time- and switching signal-dependent discontinuous Lyapunov functions, sufficient conditions are derived for the switching dynamic error system of time scheduling to maintain asymptotic stability under the average dwell time constraint.

2. The model-free smooth transition control method for aero-engine mode switching according to claim 1, characterized in that, Step 1 is described in detail below: A design method based on linear active disturbance rejection control is introduced, which can estimate and compensate for unmodeled dynamics and external disturbances in the aero-engine control system in real time. The mathematical expression of the aero-engine control system is described as follows: in, These are the state variables of an aero-engine, including the low-pressure rotor speed. High-voltage rotor speed Fuel flow and These are the dynamic characteristics of the aero-engine control system and the nonlinear functions of external disturbances, respectively. It is the measurement output of the aircraft engine control system. The coefficients are known. yes The first derivative; Within the flight envelope of an aero-engine, its operating state changes continuously over time and is divided into multiple subsystems. Indicates the time of the aircraft engine control system Operating mode; ,in It is the number of subsystems. , Represents positive integers. Represents the set of nonnegative real numbers; Linear active disturbance rejection control treats the coupling effects between channels and external disturbances caused by the environment as the total disturbance of the aero-engine control system. Each channel independently estimates the total disturbance in real time through time-scheduled switching of the extended state observer; the total disturbance is defined as... The mathematical expression for the aero-engine control system is rewritten as follows: in, , , , express The derivative of, and let The amplitude is finite, that is, there exists a... Makes all have .

3. The model-free smooth transition control method for aero-engine mode switching according to claim 2, characterized in that, Step 2 is described in detail below: The time-scheduled linear active disturbance rejection controller (ADRC) is referred to as the time-scheduled linear ADRC. During the operating time interval... The time-scheduled linear active disturbance rejection controller is divided into a transient controller and a steady-state controller; wherein, the operating time interval and The internal controller is a transition controller, while the running time interval... The internal controller is a steady-state controller; among which, This represents the segmentation score of the transition controller in the i-th subsystem. This represents a pre-defined fixed period; furthermore, by considering the actual operating time interval of the aero-engine control system, a time scheduling parameter is introduced. To dynamically adjust the maximum number of updates within the operating time interval of the time-scheduled linear active disturbance rejection controller. Represents a positive integer; the time scheduling parameters of the time-scheduled linear active disturbance rejection controller are calculated using the following equations: The transient controller and steady-state controller of the time-scheduled linear active disturbance rejection controller are both composed of two parts: a time-scheduled switching extended state observer and a time-scheduled switching feedback control law. (2.1) Time-scheduled switching extended state observer: in, and respectively and The estimated value, Indicates time The high-pressure rotor speed of the aircraft engine, Indicates time The total disturbance; and The time-scheduled switching extended state observer's first and second order gain parameters are adjusted according to the following rules: ,and It is the bandwidth of the linear-time scheduling switching extended state observer. This represents the control input gain coefficient. Indicates control input; (2.2) Time scheduling switching feedback control law: in, It is the reference input signal. It is the gain of the active disturbance rejection controller; The time-scheduled linear disturbance-free switching active disturbance rejection controller not only maintains the continuity of control gain but also takes into account the impact of the actual switching signal; the operating time interval of the transition controller is dynamically adjusted according to the actual operating time of the aero-engine control system; for the operating time interval... The signal for switching operating modes is and ,in Proposed for use from subsystem Switch to subsystem The time-scheduled variable continuous gain switching controller has the following formula: in, Is The time scheduling parameters at any given moment For the subsystem before switching At the switching time The controller gain; The time-scheduling switching feedback control law is rewritten as follows: make , and These represent tracking error, state estimation error, and total disturbance estimation error, respectively. in, The derivative of r(t); Represents time t The derivative; Therefore, the dynamic error system for time scheduling switching is represented as: Where the state vector , , It is a column vector; 。 4. The model-free smooth transition control method for aero-engine mode switching according to claim 3, characterized in that, Step 3 is described in detail below: By performing stability analysis on the switching dynamic error system of time scheduling and designing a time- and switching signal-dependent discontinuous Lyapunov function, sufficient conditions for the switching dynamic error system of time scheduling to remain asymptotically stable under the average dwell time constraint are derived. First, construct the time- and switching signal-dependent multidiscontinuous Lyapunov functions: Based on this, define a matrix: in , This indicates that the diagonal matrix has parameters. ; and yes elements, It is a piecewise linear interpolation function, specifically in the form of: (3.1) The running time interval is hour, This represents the decay rate of the Lyapunov function of the transient controller; in, The derivatives of the constructed time- and switching signal-dependent discontinuous Lyapunov functions; This indicates that under subsystem i, the scheduling parameters are based on the current time. The system matrix in the switching dynamic error system determined by the selected values; From the above equation, we get: in, Represents symmetrical elements. , Further results were obtained: in, , ; according to ,get: Similarly: If the following conditions are met: Then it satisfies ; (3.2) The running time interval is hour: Its proof method and running time range are as follows: The times are the same, therefore the following condition is satisfied: Then it is guaranteed that: (3.3) The running time interval is hour, The decay rate of the multidiscontinuous Lyapunov function representing the transient controller is expressed by: To satisfy ; On the other hand, consider the subsystem j At any moment Switch to subsystem i and subsystem j During the running time interval Internal execution; definition Running time interval The length; then, if ,inequality , Indicates the first j Each subsystem transition controller is divided into fractions. It is a constant that satisfies This is used to describe the jump in the Lyapunov function value at the switching moment. Indicates rounding down; This means: Combination By definition, we obtain: in, For the first The left limit of each switching moment, that is, the instant before the switching occurs; definition Running time range The transient controller runtime is set as follows: Analyze the overall non-incremental characteristics; in, Indicates the current time The Dolyapnov function of the activated subsystem, To indicate the system at the switching moment The state value, (·) is an exponential function. It is in the interval The total number of handovers; n is an integer variable used only to accumulate the time difference between adjacent handover moments; Scaling the above formula: in, , To represent the system from the initial time... up to the current moment Total runtime; This refers to the average length of stay. For jitter boundaries; For a switching signal and If a positive number exists ,and This makes the following equation true: Then, when the average length of stay meets the following conditions: hour, Converging to zero, the dynamic error system of time scheduling switching is affected by disturbances. The situation is gradually stabilizing.

Images