RU2779250C2 - Mixed system for aviation engine control and corresponding regulation method - Google Patents

Abstract

FIELD: computer technology.

SUBSTANCE: present technical solution relates to the field of computer technology for aviation. A system for aviation engine control contains a tracking contour, which receives set values of engine operating parameters at an input, and which controls the system together with a feedback contour according to these operating parameters. In this case, the tracking contour contains a decentralized control unit with one-dimensional regulators, a control unit with status feedback, embedded in the tracking contour. The specified control unit receives output data of the decentralized control unit with one-dimensional regulators at an input, and it is a feedback contour between engine operating parameters and outputs of the specified decentralized control unit, while the feedback contour issues engine control parameters. In this case, the control unit with status feedback is made with the possibility of decoupling of the specified operating parameters, while one-dimensional regulators of the decentralized control unit are made with the possibility of automatic regulation of the specified operating parameters according to set values of these parameters.

EFFECT: provision of full decoupling with simultaneous preservation of a possibility of simple corrector regulation.

12 cl, 7 dwg

Description

Technical field

In order to optimize the performance of aircraft (gas turbine) engines over the entire flight area, one possibility is to add "variable geometry elements" (unloader valves, variable angle blades or vanes, variable cross-section nozzle flaps, etc.). In general, variable geometry bodies are assemblies equipped with movable elements whose position can be controlled to change parameters such as the passage of a gas flow, and which thus allow degrees of freedom to be obtained in controlling a gas turbine engine. Usually they are controlled by a computing device and their actions must be coordinated in order to be optimal in terms of efficiency and remain within the acceptable limits of controllability and safety of operation of a gas turbine engine (in particular with regard to pressure, temperature, speed conditions, torques ...).

One usually speaks of a multidimensional system when the system has multiple inputs and/or multiple outputs. Classical one-dimensional architecture and control strategies are not suitable for these systems due to interactions between different variables.

Turboprop case

A turboprop engine classically comprises a gas generator (turbine) driving a variable pitch propeller.

The main task of the turboprop control system is to provide the required thrust while maintaining a constant propeller speed. From the point of view of control, a turboprop engine can be considered as a multidimensional system having:

- two input quantities:

- fuel consumption, hereinafter referred to as WF,

is the propeller pitch (also called pitch angle), hereinafter referred to as β;

- two output quantities:

- propeller power, hereinafter referred to as SHP,

- the speed of rotation of the propeller, hereinafter referred to as XNP.

Generally, the power SHP is controlled by affecting the fuel consumption WF, while the propeller speed XNP is controlled by affecting the pitch β of the propeller. The XNP speed is automatically adjusted around a number of horizontal speeds determined by the flight conditions and the state of the turboprop.

The main control problem is that a request to change the power SHP _ref results in an undesirable change in the propeller speed XNP. Similarly, a request to change the set propeller speed XNP _ref affects the gas generator power SHP. This is due to the fact that each of the commands acts on each of the outputs.

These changes adversely affect the turboprop engine. In particular, they lead to significant overtorques, which affect the fatigue wear of the components, in particular the gearbox.

Other application examples

All of the above also applies to motor control in other application examples.

For example, this refers to the control of an engine containing non-faired counter-rotating twin propellers (or “Open Rotor” in the terminology commonly used in the field), which is an example similar to a turboprop in terms of control: the system uses three inputs (fuel flow and corresponding pitches of two counter-rotating propellers) to control three outputs (rotation speed of the low pressure chamber and rotation speeds of each of the two propellers).

This also applies to turbojet engines, in which, in addition to the fuel consumption, the section of the jet nozzle can also be controlled. This additional degree of freedom basically allows you to influence the speed of the release of gas and, therefore, thrust.

Variable geometry elements can also be integrated into turbojet engines. Although the setpoints for variable geometries and for fuel flow are developed in concert, their control systems (servo valve currents) are synthesized independently and may affect some common operating parameters such as rotational speeds or compressor performance curve.

The main tasks in the synthesis of control rules

The main tasks in the synthesis of control rules for these systems are:

- Compliance with the specifications given by the technical requirements (response time, exceedance, stability margins),

- Limitation of interactions between different variables: changing the set value of one of the quantities should lead to the smallest possible deviation on the other quantities,

- Stability of control to deviations and modeling errors.

In addition to efficiency objectives, control rules should be easily adjustable and of reasonable complexity to facilitate regulation, correction during testing, and application in computing devices.

Existing Decoupled Multidimensional Control Solutions

In these solutions, the synthesis of control rules is carried out on the basis of a linear model identified around the operating points. The following diagrams of known systems are presented with linear models of a turboprop engine. These models are designated G if the form of the transition matrix is used. When the state form is used, the turboprop model consists of the state, control, and output matrices A, B, and C (while in the considered cases of engine control, the direct-acting matrix D is zero).

As shown in FIG. 1, the so-called centralized solutions of the type H ₂ , H _∞ are known (as the solution described in the document G. Zames, Feedback and Optimal Sensitivity Model Reference Transformations, Multiplicative Seminorms, and Approximations, IEEE Transactions on Automatic Control, vol. 26, issue n ° 4, 1981).

These approaches make it possible to synthesize a multidimensional corrector 11 by directly taking into account the power SHP and the speed XNP of the propeller.

Although these solutions are relatively efficient, their development requires complex calculations, and the resulting equalizers are usually complex. In addition, these control rules are not intuitive, which creates an obvious problem for correcting adjustments during motor testing.

As shown in FIG. 2, variants of state feedback control (such as the pole placement method, quadratic linear control method) are also known. These methods consist of adding a feedback that includes feedback on state 12. As a rule, terms of the integral action 13 are added to the feedback on state 12 in order to achieve good performance.

Since these methods do not provide decoupling, the interactions between the power loops of the SHP and the XNP propeller remain too significant. In addition, state-feedback motor control is virtually uncorrectable on the test bench.

As shown in FIG. 3, state feedback decoupling approaches are known (eg, such as Falbe-Volovich synthesis (described in Multivariate Control and Evaluation, Ostertag, 2006) or full modal synthesis). These methods consist of adding feedback including state feedback 14 and static compensation M 14b to decouple the system.

The advantage of these methods is that they provide complete decoupling using relatively simple equalizers.

However, these methods are not compatible with adding integral actions. Indeed, these methods are based on moving a limited number of system dependent poles, and adding integrators and hence unstable poles results in moving the latter instead of the system poles. The efficiency achieved is insufficient, in particular in terms of interference elimination. In addition, in practice it is difficult to correct control systems with feedback on the condition.

As shown in FIG. 4, there are also known frequency decoupling methods (such as the pseudo-diagonalization method, the eigenvalue and singular value decomposition method, the simplified decoupling scheme or the ideal decoupling scheme method), which allow the system to be decoupled using compensators 15 before using one-dimensional controllers 16 PID type, allowing automatic control of the system.

The advantage of these methods is that the stages of synthesis of compensators 15 and one-dimensional controllers 16 are separate. Therefore, it is relatively easy to correct the a posteriori adjustments of the one-dimensional controllers 16.

The problem is that, as a rule, simple compensators do not provide satisfactory decoupling. Proper or ideal decoupling can only be achieved using complex compensators that are almost impossible to implement and interpolate.

Some methods consist only of using the compensation matrix 15 as offsets added to the commands to compensate for the interference of various circuits. The problem with these methods is that the offsets are difficult to calculate and that they are not always accurate enough.

The above approaches allow:

- get an average interchange, or

- get the correct outcome due to relatively complex control rules and synthesis methods.

In addition, an aircraft engine system is known from US 5,274,558 which contains:

- the servo circuit, which receives at the input the set torque value QCMD,

- rotation speed loop (Fig. 3 - speed setpoint NPSFLT).

The system also includes a decoupler 18, 35 which outputs a signal indicating the decoupled torque QDC, which allows the torque loop to be separated from the speed loop so that only the speed loop responds to abrupt changes in engine speed.

No status feedback control is provided here.

US 5001646 describes an electrical control system (“fly-by-wire”) for a helicopter.

This system contains at least one servo circuit, which receives at the input the specified values of the parameters (roll, pitch, yaw, vertical) coming from the control stick.

The feedback loop regulates the operating parameters of the motor.

This system also does not provide state feedback control.

US 5920478 describes a general dynamic control system with a servo loop.

The decoupler processes the MIMO data and transforms it into data that provides simple one-dimensional SISO processing while taking into account the complex interactions between variables.

The applied decoupling is not a state feedback type decoupling, but corresponds to the decouplings presented above with reference to FIGS. 4. As mentioned above, the disadvantage of these decoupling schemes is that they do not provide satisfactory decoupling, unless extremely complex compensators are used.

Disclosure of invention

The general object of the invention is to provide the necessary efficiency, in particular with regard to decoupling, while maintaining the possibility of adjusting the equalizers in a simple and intuitive way.

For this, an aircraft engine control system is proposed, which contains:

- at least one servo loop, which receives at the input the set values of the engine operating parameters and which controls the system together with the feedback loop according to these operating parameters, while the specified servo loop includes a decentralized control unit with one-dimensional controllers,

- at least one control unit with status feedback built into the servo loop, wherein the specified control unit receives at the input the output data of the decentralized control unit with one-dimensional controllers and is a feedback loop between the operating parameters of the engine and the outputs of the specified decentralized control unit, at the same time, the specified feedback loop outputs the engine control parameters,

wherein the control unit with status feedback is configured to decouple operating parameters, while one-dimensional controllers of decentralized control are configured to automatically control operating parameters according to the specified values of these parameters.

Decentralized control in this case should be understood as control that is not centralized, but that uses a variety of local and independent control rules for various managed subsystems.

The feedback control may include a static compensator and a state feedback corrector circuit, wherein said static compensator and said state feedback corrector circuit are configured to decouple operating parameters.

Such a system combines the advantages of state feedback decoupling solutions and frequency decoupling techniques.

In particular, the decoupling circuit and regulators are separated.

Since the decoupling scheme is implemented using a state feedback decoupling approach, it is possible to achieve full decoupling with a very simple state equalizer and a compensation matrix M, while maintaining, unlike classical decoupling methods, the main dynamic factors of the method (which is expressed in the matrix by values, equal to 0, on off-diagonal elements, and non-zero values on diagonal elements).

Thus, reducing ties is very effective. In addition, decoupling state feedback consists of two simple gain matrices: state feedback and static compensation.

Preferably, the one-dimensional decentralized control controllers are proportional-integral correctors (PI correctors).

PI-correctors make it possible to obtain the required characteristics.

Their advantage lies in their simplicity, and their installations are facilitated by the decoupling. In addition, they can be easily adjusted/reset, regardless of decoupling state feedback. These adjustments/resets can be made during the design phases, during bench tests, and also during the operation of the turboprop. Indeed, as a system ages, its behavior changes. In this case, the PI equalizers can be adapted to maintain the desired performance (particularly with regard to response time and overshoot).

The subject of the invention is also a method for parameterization of the aforementioned aircraft engine control system, comprising the steps of:

- define a linear model of the engine, while the model has a transient function and state representation;

- determine the static compensator and the status feedback corrector in such a way as to decouple the states formed by the operating parameters of the regulated motor;

- define one-dimensional controllers, so as to automatically adjust the specified operating parameters to setpoints.

Thus, a mixed decoupling is implemented, in which:

- perform decoupling by means of status feedback;

- synthesize controllers (in particular, PI controllers) using a method decoupled by means of state feedback.

In particular, the condition feedback corrector and the static compensator can be configured such that the control system transient function has gains and poles corresponding to the gains and poles of the motor model transient function.

The method may further comprise the step of interpolating one-dimensional correctors depending on variable flight conditions.

The parameters of the one-dimensional equalizers can be interpolated individually by specifying a sequence of gain factors.

The method may also comprise the step of interpolating the status feedback corrector and the static compensator depending on the variable flight conditions.

The invention also relates to the use of the proposed system for various applications.

It should be noted that the solution proposed by the invention is extremely effective in the case of a system considering 2 or 3 controls and outputs, while it is possible to provide for a larger number of variables, but perhaps with less accurate results. Indeed, usually control rules are synthesized on a nominal linear model. Since this model is not ideal, the control rules do not provide the same level of efficiency on a non-linear model. The same applies to decoupling, which is known to be quite sensitive to model errors. The M and L matrices account for all transitions in the multidimensional system and may become less accurate as the number of variables increases.

In the case of controlling a variable pitch propeller engine such as a turboprop, the automatically controlled operating parameters include propeller power (SHP) and propeller speed (XNP), the controlled parameters are fuel consumption and propeller pitch.

In the case of an engine with counter-rotating twin propellers (contra-rotating twin variable-pitch propellers, not “Open Rotor” or “Contrafan”), in which the automatically controlled operating parameters include the low pressure chamber mode ( NBP) and two propeller mode (N1 and N2), and controlled parameters include fuel consumption (WF) and propeller pitch (β ₁ and β ₂ ).

To control a variable nozzle turbojet, the automatically controlled operating parameters include the low pressure chamber (NBP) mode and the position (xT) of the power cylinder or power cylinders controlling the variable nozzle, while the controlled parameters include fuel consumption (WF) and current (iTuy) of the servo valve supplying the power cylinder or power cylinders.

In the case of variable geometry turbojet control, automatically controlled operating parameters include low pressure chamber (NBP) mode and power cylinder position (xVSV and xVBV) (other operating parameters such as power steering performance curve can also be provided) , and controlled parameters include fuel consumption (WF) and control currents (iVSV and iVBV) of the servo valves.

Brief description of the drawings

Other features and advantages of the invention will become more apparent from the following description, which is presented by way of illustrative and non-limiting example only, with reference to the accompanying figures, in which:

Fig. 1 is a diagram of a known centralized control system.

Fig. 2 is a diagram of a known state feedback control system.

Fig. 3 is a diagram of a known state feedback decoupling system.

Fig. 4 schematically illustrates a known method of frequency decoupling.

Fig. 5 shows a mixed control strategy system according to an exemplary embodiment of the invention.

Fig. 6 shows the control of the system shown in FIG. 5, with status feedback.

Fig. 7 is an example of the application of the invention to the control of a turboprop.

Description of embodiments of the invention

In the example shown in FIG. 5, the case of two setpoints, commands and outputs is considered, and this architecture and control method is further generalized to N setpoints, commands and outputs.

The architecture of this control system 100 includes:

- decentralized control circuit 101 with compensators;

- a decoupling system 103 with status feedback built into said loop 101.

The loop 101 receives N given values Yref and contains a decentralized control unit 102 whose outputs feed the state feedback decoupling system 103 and whose equalizers are selected so that the loop 101 can provide the automatic control function of the method.

Status feedback decoupling

The state feedback decoupling system is shown more specifically in FIG. 6 and contains:

a static compensator 104 configured to decouple the setpoints at the output of the decentralized control unit 101 (pre-filter matrix M (or compensation matrix));

- status feedback corrector 105 (state feedback matrix L).

In what follows, it is assumed that the following hypothetical conditions are met:

- The method is linear.

- The number of commands is equal to the number of controlled outputs. We are talking about the classical hypothesis for decoupling (and in any case it is impossible to decouple the system if you do not have the necessary number of degrees of freedom).

- The matrix of direct action is zero. This is the classical hypothesis followed by most state feedback control systems (commands do not act instantly on outputs).

- The control matrix B is irreversible.

- The method does not contain an unstable zero (as in the case of other state feedback decoupling methods, an unstable zero would lead to an unstable pole in the solution).

Let us also consider the case when the number of system states is equal to the number of outputs (all considered states are measured). In this case, you can adjust A and B to treat the output matrix C as an identity matrix.

However, this hypothesis can be circumvented. In particular, in the case when the system has a number of states that exceeds the number of inputs/outputs, an observation algorithm can be added.

Considering the above hypothetical conditions, the output matrix C of the circuit in FIG. 3 equals identity by assumption, so:

и

and

The state representation d can be written as follows:

: столбец, представляющий N переменных состояния

: column representing N state variables

: столбец, представляющий N команд

: column representing N commands

: столбец, представляющий N выходов

: column representing N outputs

: Матрица состояния

: Status Matrix

: Матрица управления

: Matrix control

The system under consideration can be fully represented in the form of the following expression (1). Then you can use the notation of expression (2).

Therefore, the control exercised by the decoupling system shown in FIG. 6 is such that:

As for L and M, they are chosen in such a way that:

where the parameters p _i and G _i are scalar values.

Indeed, such a control provides an ideal decoupling:

, зависит только от его соответствующего элемента в векторе х (например,

должен зависеть только от

, но не от других элементов х),- each element included in

, depends only on its corresponding element in the vector x (for example,

should only depend on

, but not from other elements x),

, влияет только соответствующий ему элемент в векторе u (например, на

влияет только

, но не другие элементы u).- for each element included in

, affects only the element corresponding to it in the vector u (for example, on

affects only

, but not other elements of u).

Indeed, with such matrices L and M:

и

.We obtain (7) by combining expressions (3), (5) and (6). Passing to (8) and (9) only uses the fact that

and

.

Passing to the Laplace region, we get:

Thus, the responses of the system are decoupled and correspond to first-order transient functions with poles p _i and gains G _i .

The values of the gains G _i and the poles p _i are determined by approximating the diagonal terms of the method using the first orders of equation (10). Various simplification methods make it possible to choose the gains G _i and poles p _i based on the diagonal transitions of the system.

For example, model simplification techniques such as the clipping method are used to do this, and the frequency responses of the original transitions and the frequency responses of the reduced transitions are compared while trying to maintain the static gain and cutoff frequency.

In particular, the gains G _i and the poles p _i can be chosen such that the static gains and cutoff frequencies are similar.

It should be noted that with the proposed control scheme, the resulting decoupling is complete, while the installation of automatic control is carried out not on the state feedback loop 103, but on the correctors/compensators of the loop 101.

Thus, in contrast to the so-called Falbe-Volovich method [Multidimensional control and evaluation, Ostertag, 2006], the task of a state feedback system is not a method of automatic control, but a method of its decoupling.

This allows the corrector gains (see below) to best preserve the dynamics of the transitions.

One-dimensional correctors

The correctors of the decentralized control unit 102 are selected in such a way as to correct the necessary automatic control based on the decoupled system.

In particular, when these equalizers are PI equalizers, their relatively simple structure greatly facilitates parameter settings. Indeed, the shape of the PI-correctors makes it possible to retain a certain physical meaning, which makes parameter settings more intuitive. They can be adjusted independently of the decoupling during bench tests.

If a different decoupling method is used and if the decoupling is not complete, the so-called multiloop synthesis methods (detuning methods, cyclic methods...) must be used. Such methods make it possible to take into account the interactions between the SHP and XNP circuits during the synthesis of correctors, but they are more complex than the classical control methods.

Since the state-feedback decoupling is complete, classical parameter setting methods can be used to tune the equalizers. These methods are easy to use as settings can be made independently to automatically adjust system outputs.

In particular, the PID-IMC (Internal Model Control) method can be used to make these adjustments based on the decoupled process transients and specifications expressed as first order transients. Adjustments of PI controllers are relatively simple and can be done automatically with this method, based on decoupled process transients and specifications.

However, other 1D correctors and other tuning methods can be used instead of the PID-IMC method.

Parameterization

The control system is parameterized in accordance with the method of parameterization, containing the steps in which:

- E1 determine the model of the considered mechanical system of the engine;

- E1 determine the corrector 105 feedback on the state and the static compensator 104 in such a way as to decouple the controlled parameters of the outputs, while maintaining the basic dynamic characteristics of the system;

- E3 determine the correctors block 102 decentralized control (for example, one-dimensional PI-correctors) to automatically adjust the output parameters in accordance with the selected rule.

Optionally, the method may comprise the step E4 of interpolating the equalizers of the decentralized control unit 102, the status feedback corrector 105, and the static compensator 104.

Application for turboprop

In FIG. 7 shows an application for controlling a turboprop engine T.

The control system shown in this figure is given input propeller power SHPref and propeller speed XNPref.

These setpoints are generated, for example, by a control unit (not shown) which calculates said power and speed setpoints based on the thrust setpoint that comes from the control device 18, typically a control stick.

This system outputs the propeller power SHP and the propeller speed XNP of the turboprop, which are measured by the SHP propeller power sensor 106 and the XNP propeller speed sensor 107.

The system comprises two common control loops 101a and 101b, of which one (101a) automatically sets the power SHP to the power setpoint SHPref, and the other (101b) automatically sets the propeller speed XNP to the setpoint propeller speed XNPref.

At the input of the common power loop SHP 101a is a differentiator 108a, and at the input of the common propeller speed loop XNP 101b is a differentiator 108b.

Differentiators 109a and 109b also receive propeller power SHP and rotational speed XNP as inputs and redirect them to the feedback inputs of two loops 101a and 101b.

Each of these two circuits 101a, 101b contains, in turn, a decentralized control unit 110a, 110b and two decoupling circuits 103a, 103b, respectively fed through the outputs of the decentralized control units 101a, 101b.

Each decentralized control unit 101a, 101b is input with a loop error signal to which it corresponds.

The decentralized control unit 110a includes a one-dimensional SHP power equalizer configured to automatically adjust the SHP power to a set value SHPref, while the decentralized control unit 110b includes a one-dimensional XNP rate corrector configured to automatically adjust the XNP rate to a set value XNPref.

Decoupling loops 103a, 103b include a status feedback corrector 105 configured to decouple propeller power SHP and propeller speed XNP of turboprop 1 in terms of system states.

Said equalizer 105 receives as input the output of differentiators 109a and 109b.

The outputs of the corrector 105 form the feedback inputs of said circuits 103a and 103b, which receive at their other input the output of the static compensator 104.

Said circuits 103a, 103b also respectively comprise an integrator 111a located before the fuel rate control input WF of the turboprop T, and an integrator 111b located at the transition of the propeller pitch control input β of the turboprop T.

These two integrators 111a and 111b generate respectively the fuel flow command WF and the propeller pitch command β for the turboprop T.

A certain number of characteristic operating points are determined. These characteristic operating points are characterized by flight conditions (Mach number, flight altitude) and turboprop engine conditions.

During the design phase, identification scenarios are modeled for each of these characteristic operating points based on a non-linear thermodynamic model of the engine. The control used for these identification scenarios is a sequence of desynchronized control stages with white noise.

The results of these simulations are then used to characterize the behavior of the turboprop and define a linear turboprop model that models the operation of the turboprop and whose inputs are fuel consumption WF and propeller pitch β and whose outputs are propeller power SHP and propeller speed XNP.

After analyzing the results, a turboprop order 2 model is selected. Thus, the turboprop model has two state variables, with the order of the system being equal to the number of state variables.

Since the number of state variables is equal to both the number of inputs and the number of outputs of the turboprop model, it is possible to directly link the state variables to the outputs of the turboprop model. Thus, the turboprop state variables are propeller power SHP and propeller speed XNP.

In steps E2 and E3, a computer numerical simulation of the operation of the turboprop is applied using the turboprop engine model determined in step E1.

The analysis of these data led to the need to define order 2 models without direct action.

The transition matrices of the identified linear systems have the form (11) as a state and (12) as a transition matrix. After analyzing the zeros and matrices B of linear systems at all operating points, it turns out that the former are stable and that the matrices B are irreversible.

Thus, a mixed decentralized management strategy can be applied.

Based on this conclusion, the parameter determination described above is applied for each of the identified systems.

At the first stage (E1) there is an automatic search for the main dynamic conditions of the diagonal transitions of the turboprop on each of the identified models. This can happen automatically through certain model simplification techniques, such as truncation of the balanced base. This makes it possible to find first-order transitions that bring together diagonal transitions consisting of two poles and zero. From here G1, G2, p1 and p2 are deduced. Preferably, the gains Gi and poles pi are chosen such that the static gain of the simplified transition frequency responses does not deviate from the static frequency response gain of the initial transitions and that the cutoff frequency of the simplified transition frequency responses does not deviate from the cutoff frequency of the initial transition frequency responses.

At the second stage (E2), the matrices M and L are analytically determined at each operating point using (10).

In the third step (E3), the PI correctors for the SHP power loops and the XNP propeller mode are determined. This is also carried out analytically using the FID-IMC method.

The M, L matrices and PI correctors change over the flight region to compensate for system variability. To maximize control efficiency, control rules are interpolated depending on flight conditions (Mach number, flight altitude, turboprop condition).

Thus, the PI equalizers are interpolated to improve system performance (static error, response time, overflow) and the M and L matrices in order to maintain optimum decoupling.

The interpolation of equalizers makes it possible to increase the efficiency of the control system (in particular with regard to static error, response time and overflow).

The interpolation of the status feedback corrector 105 and the static compensator 104 maintains optimum decoupling over the entire flight area.

In particular, the parameters of one-dimensional equalizers can be interpolated individually by ordering the sequence of gain factors.

Gain sequencing consists of defining a family of linear systems, approaching a non-linear system at a given number of operating points, and proposing control rules in each of the respective regions of the state space to eventually implement the overall control rule.

In particular, if the 1D equalizers are PI equalizers, the gain interpolation is simple since it is a sum of different actions weighted by the gains. Similarly, the condition feedback corrector 105 and the static compensator 104 are 2×2 gain matrices L and M, so it is relatively easy to interpolate the coefficients of these matrices.

An incremental algorithm can be applied to smooth commands during control rule interpolation and provide smoother transitions when multiple control loops are used in parallel. In this way, the setpoints and outputs are differentiated (differentiators 108a, 108b and 109a, 109b), which allows error increments to be handled, with the control increments being subsequently integrated.

The control rules allow for optimal responses over the entire flight area, while the control rules are relatively simple and easy to adjust.

Application to other systems

For other applications, the decision steps are the same as for a turboprop.

и

. Что касается выходных данных, то они включают в себя режим корпуса низкого давления N_BP и режимы двух воздушных винтов N₁ и N₂, как показано ниже:In the case of a powerplant containing twin contra-rotating propellers not hooded “Open Rotor” or hooded “Contrafan”, commands include WF fuel consumption and propeller pitches

and

. As for the output, it includes N _BP low pressure case mode and N ₁ and N ₂ twin propeller modes as shown below:

The system, linearized at various operating points, is expressed by a full (all terms are non-zero) 3×3 transition matrix. Various hypothetical conditions are met, which makes it possible to obtain the matrices L and M described in (10).

In the case of turbojet engines with a variable nozzle section, the control commands include the fuel consumption WF and the current of the servo valve i _Tuy , which feeds the power cylinder, which allows the variable nozzle section to be changed. As for the output data, they include the N _BP low pressure chamber mode and the consolidated position of the X _Tuy power cylinders:

The system is linearized on a smaller number of operating points. The linearized system corresponds to a 2×2 matrix, but the WF→x _Tuy transition is zero (changes in fuel flow have no direct effect on nozzle position). Thus, we are talking about minimizing another binding term, namely i _Tuy →N _BP . This does not change the hypothetical conditions, and formula (10) can be applied.

In the case of turbojet engines with variable geometry elements, the control commands include the fuel consumption WF and the control currents of the servo valves i _VSV and i _VBV , providing power to the power cylinders, respectively, to optimize the angle of the stators (VSV from Variable Stator Vanes - adjustable stator blades) of the compressor high pressure and for modulating the opening bias of unloading valves (VBV stands for Variable Bleed Valves). Automatically controlled outputs include N _BP low pressure casing mode and x _VSV and x _VBV power cylinder positions. As regards the setpoints of these drives, they are perfectly synchronized.

The system, linearized at various operating points, is described by a 3×3 transition matrix. The predominant terms are the diagonal terms, and the transitions WF→x _VSV and WF→x _VBV are zero. On the other hand, other coupling terms are non-zero (the VBV action affects the low pressure mode directly, and the VSV action also affects it through the high pressure mode). Here, too, the stated solution can be applied.

Claims (18)

1. An aircraft engine control system, characterized in that it contains: - at least one servo loop, which receives at the input the set values of the engine operating parameters and which controls the system together with the feedback loop according to these operating parameters, while the specified servo loop contains a decentralized control unit with one-dimensional controllers, - at least one control unit with status feedback built into the servo loop, wherein the specified control unit receives at the input the output data of the decentralized control unit with one-dimensional controllers and is a feedback loop between the operating parameters of the engine and the outputs of the specified decentralized control unit, at the same time, the specified feedback loop outputs the engine control parameters, wherein the control unit with status feedback is configured to decouple said operating parameters, wherein the one-dimensional controllers of the decentralized control unit are configured to automatically control said operating parameters according to the specified values of these parameters. 2. The system of claim 1, wherein the state feedback control unit comprises a static compensator and a state feedback corrector circuit, wherein said static compensator and said state feedback corrector circuit are configured to decouple operating parameters. 3. The system according to claim 1 or 2, in which the one-dimensional controllers of the decentralized control unit are proportional-integral correctors. 4. The method of parameterization of the aircraft engine control system according to any one of paragraphs. 1-3, containing the steps in which: - define a linear model of the engine, while the model has a transient function and state representation; - determine the static compensator and the status feedback corrector in such a way as to decouple the states formed by the automatically controlled operating parameters of the engine; - define one-dimensional controllers in such a way as to automatically adjust the specified operating parameters to specified values. 5. The control system parameterization method according to the previous paragraph, in which the state feedback corrector and the static compensator are configured in such a way that the control system transient function has gains and poles corresponding to the gains and poles of the motor model transient function. 6. The method of parameterizing the control system according to claim 4 or 5, additionally containing the step of interpolating one-dimensional correctors depending on the variable flight conditions. 7. The control system parameterization method according to claim 6, wherein the parameters of the one-dimensional correctors are interpolated individually by specifying a sequence of gain factors. 8. The method of parameterization of the control system according to any one of paragraphs. 4-7, further comprising the step of interpolating the status feedback corrector and the static compensator depending on the flight conditions variables (C1, C2, C3, ...). 9. The use of the system according to any one of paragraphs. 1-3 to control a variable pitch propeller engine such as a turboprop, where the automatically controlled operating parameters include propeller power (SHP) and propeller speed (XNP), while the controlled parameters include fuel consumption and propeller pitch. . 10. The use of the system according to any one of paragraphs. 1-3 to control an engine having counter-rotating, variable-pitch twin propellers, the auto-adjustable operating parameters include low pressure chamber (NBP) mode and dual propeller modes (N1 and N2), and the controlled parameters include includes fuel consumption (WF) and propeller pitch (β ₁ and β ₂ ). 11. The use of the system according to any one of paragraphs. 1-3 to control a turbojet engine with a variable nozzle section, while the adjustable operating parameters include the mode of the low pressure chamber (NBP) and the position (xT) of the power cylinder or power cylinders that control the set nozzle section, while the control parameters include fuel consumption (WF) and current (iTuy) of the servo valve supplying the power cylinder or power cylinders. 12. The use of the system according to any one of paragraphs. 1-3 to control a turbojet engine with elements having variable geometry, while automatically adjustable operating parameters include low pressure chamber mode (NBP) and power cylinder positions (xVSV and xVBV), control parameters include fuel consumption (WF) and control currents (iVSV and iVBV) for servo valves.

Images