Rolling optimization predictive closed-loop controller and control system

Abstract

The present application provides a rolling optimization prediction closed-loop controller and an engine rolling optimization prediction closed-loop control system. The rolling optimization prediction closed-loop controller comprises an engine information calculation module, a thrust prediction closed-loop controller, a steady-state controller, a transition-state controller, a start-state controller and a fuel control amount selector. The engine information calculation module is used to determine the required engine dynamic model information and engine state information. The thrust prediction closed-loop controller is used to send the obtained first calculation result to the fuel control amount selector. The steady-state controller is used to send the obtained second calculation result to the fuel control amount selector. The transition-state controller is used to send the obtained third calculation result to the fuel control amount selector. The start-state controller is used to send the obtained fourth calculation result to the fuel control amount selector. The fuel control amount selector is used to output the corresponding fuel control amount, so as to output the corresponding fuel instruction.

Description

Technical Field

[0001] This invention relates to the field of engine thrust control, and more particularly to a rolling optimization predictive closed-loop controller and an engine rolling optimization predictive closed-loop control system. Background Technology

[0002] With the advent of full authority digital technology in the engineering application field of aero engines, the demand for engine safety protection has been continuously increasing, especially the thrust control safety margin during the dynamic operation of aero engines. Major aircraft engine manufacturers and related research institutions around the world have invested a lot of research and development resources in this area.

[0003] Currently, there are very few reports on major engine OEMs (Original Equipment Manufacturers) planning to directly control engine thrust using an onboard controller in a closed-loop manner. In engineering practice, the relationship between thrust and speed is typically calibrated on a test bench (ground, flight, etc.), and a converted tachometer and thrust relationship curve are used to control the converted speed to achieve thrust control. This approach requires test calibration for various engine operating scenarios, has high development costs, and poor dynamic adaptive capabilities.

[0004] Furthermore, Chinese invention CN201710931188.5 proposes an energy management method for hybrid-powered ships based on model predictive control (MMC). It selects a Markov model to predict the power demand of the hybrid-powered ship and, combined with the principles of MMC, uses dynamic programming to optimize power allocation within the prediction time domain with the objective of minimizing fuel consumption. However, this method suffers from the problem of cumbersome theoretical solution steps. Chinese invention CN201711134813.X discloses an algorithm for estimating the thrust of aero-engines using an adaptive RBF neural network, proposing an adaptive RBF neural network based on particle swarm optimization (PSO). In the improved PSO algorithm, a local optimum is set with the same number of hidden layer node types as the number of hidden layer node types. This method also suffers from the problem of cumbersome theoretical solution steps.

[0005] In order to overcome the above-mentioned defects in the existing technology, there is an urgent need in the field for an improved rolling optimization prediction closed-loop controller to simplify the solution process of engine dynamic thrust prediction control, improve the performance of thrust prediction, and enhance the safety margin of engine thrust control. Summary of the Invention

[0006] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed descriptions that follow.

[0007] To overcome the aforementioned shortcomings of existing technologies, this invention provides a rolling optimization predictive closed-loop controller and an engine rolling optimization predictive closed-loop control system. This controller can achieve the functions of traditional steady-state, transient, and dynamic controllers, thereby realizing closed-loop control for dynamic engine thrust prediction and simplifying the controller architecture. Furthermore, the fuel control selector utilizes a neural network model to calculate fuel control quantities, simplifying the solution process for dynamic engine thrust predictive control and improving thrust prediction performance, thus enhancing the safety margin of engine thrust control.

[0008] Specifically, the rolling optimization prediction closed-loop controller provided according to the first aspect of the present invention includes an engine information calculation module, a thrust prediction closed-loop controller, a steady-state controller, a transient controller, a dynamic controller, and a fuel control quantity selector. The engine information calculation module is used to determine the required engine dynamic model information and engine state information corresponding to the engine throttle angle signal and the thrust mode signal based on the engine throttle lever angle signal and the thrust mode signal. The engine dynamic information includes at least the engine temperature and pressure. The thrust prediction closed-loop controller is used to calculate the fuel control quantity required for the thrust prediction closed loop based on the engine dynamic model information and the first engine state information provided by the engine information calculation module, and sends the obtained first calculation result to the fuel control quantity selector. The steady-state controller is used to calculate the engine steady-state fuel control quantity required for the speed closed loop based on the engine dynamic model information and the second engine state information provided by the engine information calculation module, and sends the obtained second calculation result to the fuel control quantity selector. The transient controller is used to calculate the engine acceleration fuel control quantity and engine deceleration fuel control quantity required for acceleration control based on the engine dynamic model information and third engine state information provided by the engine information calculation module, and sends the obtained third calculation result to the fuel control quantity selector. The starting dynamic controller is used to calculate the engine ignition fuel control quantity required for starting control based on the engine dynamic model information and fourth engine state information provided by the engine information calculation module, and sends the obtained fourth calculation result to the fuel control quantity selector. The fuel control quantity selector is used to input the first calculation result, second calculation result, third calculation result, and / or fourth calculation result into a pre-trained neural network model to output the corresponding fuel control quantity, thereby outputting the corresponding fuel command.

[0009] Furthermore, in some embodiments of the present invention, the rolling optimization prediction closed-loop controller further includes a fault diagnostic unit. The fault diagnostic unit is used to receive abnormal signals from the engine information calculation module, the thrust prediction closed-loop controller, the steady-state controller, the transient controller, the engine dynamic controller, and / or the fuel control quantity selector, for signal anomaly diagnosis; and / or the fault diagnostic unit is used to detect the hardware status of the engine for hardware anomaly diagnosis.

[0010] Furthermore, in some embodiments of the present invention, the step of training the neural network model includes: acquiring historical operating data of the engine. The historical operating data includes historical state information of the engine and corresponding fuel control quantities; constructing a neural network model to be trained; and training the neural network model based on the historical operating data.

[0011] Furthermore, in some embodiments of the present invention, the engine information calculation module includes a thrust manager, a signal processor, and a controller coefficient scheduler. The thrust manager calculates the desired engine thrust command based on the engine throttle angle signal and thrust mode signal, and outputs it to the thrust prediction closed-loop controller. The signal processor calculates the actual physical value of the converted signal based on the acquired raw sensor signals, and outputs it to the thrust prediction closed-loop controller, the steady-state controller, the transient controller, and the dynamic controller. The controller coefficient scheduler receives the turbofan engine high-pressure rotor converted speed signal sent by the signal processor, and, in conjunction with stored offline acquired information, determines the engine dynamic model information required by the thrust prediction closed-loop controller.

[0012] Furthermore, in some embodiments of the present invention, the engine dynamic model is as follows:

[0013] in, The dynamic equation for the engine is given. The state of motion of the dynamic equation is denoted as . is the input quantity for the dynamic equation. The loyabi of the engine. These are linear coefficients. This is the gain coefficient. This refers to the static pressure at the outlet of the high-pressure compressor. The operating ambient temperature of the engine. Let be the expected operating thrust of the engine. This refers to the actual thrust generated by the engine during operation. This refers to the fuel control quantity.

[0014] Further, in some embodiments of the present invention, the thrust prediction closed-loop controller includes an initial value calculation module, a discrete optimization matrix calculation module, an initial control law calculation module, a control law calculation module, a sampled value update calculation module, a Jacobi matrix calculation module, a control law change rate calculation module, and a recursive least squares calculation module. The initial value calculation module is used to preset the initial input parameters corresponding to the variables. The discrete optimization matrix calculation module is used to calculate the optimal C matrix and H matrix of the controller. The initial control law calculation module, based on the initial input parameters provided by the initial value calculation module, and the H matrix and C matrix provided by the discrete optimization matrix calculation module, and based on the recursive least squares calculation module, calculates the approximate solution of the initial control law in the least squares sense, and then outputs the obtained calculation result to the control law calculation module. The control law calculation module, based on the initial control law and the control law change rate, calculates the fuel control quantity per unit time for each cycle, outputs the obtained calculation result to the sampled value update calculation module, and sends the final fuel control quantity result to the fuel control quantity selector. The sampling value update calculation module updates the sampling value of the fuel control quantity for each unit of time according to the calculation results provided by the control law calculation module, and outputs the obtained calculation results to the Jacobian matrix calculation module. The Jacobian matrix calculation module updates the Jacobian matrix for each unit of time according to the initial input parameters provided by the initial value calculation module, and outputs the obtained calculation results to the control law change rate calculation module. The control law change rate calculation module calculates the approximate solution of the control law change rate in the least squares sense based on the initial input parameters provided by the initial value calculation module and the recursive least squares calculation module, and then outputs the obtained calculation results to the control law calculation module. The recursive least squares calculation module solves the least squares equation of the input equation based on the round-robin transformation method.

[0015] Furthermore, in some embodiments of the present invention, the rolling optimization prediction closed-loop controller further includes a thrust planning converter. The thrust planning converter is connected to the thrust manager and is used to perform target calculations for steady-state control based on the engine operating thrust expectation command sent by the thrust manager, to obtain a first target control speed for the low-pressure shaft of the turbofan engine and a second target control speed for the high-pressure shaft of the turbofan engine, and outputs these values ​​to the steady-state controller. And / or the thrust planning converter performs target calculations for transient control based on the engine operating thrust expectation command sent by the thrust manager, to obtain the rate of change of the first target control speed for the low-pressure shaft of the turbofan engine and the rate of change of the second target control speed for the high-pressure shaft of the turbofan engine, and outputs these values ​​to the transient controller; and / or the thrust planning converter performs target calculations for dynamic control based on the engine operating thrust expectation command sent by the thrust manager, to obtain the rate of change of the first target control speed for the low-pressure shaft of the turbofan engine and the rate of change of the second target control speed for the high-pressure shaft of the turbofan engine, and outputs these values ​​to the dynamic controller.

[0016] Furthermore, in some embodiments of the present invention, the controller coefficient scheduler is used to provide the steady-state controller with the required control parameters and existing calculation output functions. The steady-state controller is connected to the thrust planning converter and is used to calculate the engine steady-state fuel control quantity required for speed closed-loop based on the steady-state controller control parameter information provided by the controller coefficient scheduler and the second engine state information provided by the signal processor, and send the obtained calculation result to the fuel control quantity selector.

[0017] Furthermore, in some embodiments of the present invention, the controller coefficient scheduler is used to provide the transient controller with the required control parameters and existing calculation output functions. The transient controller is connected to the thrust planning converter and is used to calculate the engine acceleration fuel control amount and engine deceleration fuel control amount required for acceleration control based on the transient controller control parameter information provided by the controller coefficient scheduler and the third engine state information provided by the signal processor, and send the obtained calculation results to the fuel control amount selector.

[0018] Furthermore, in some embodiments of the present invention, the controller coefficient scheduler is used to provide the start-up dynamic controller with the required control parameters and the calculation output function of existing rules. The start-up dynamic controller is connected to the thrust planning converter and is used to calculate the engine start-up fuel control amount required for start-up control based on the start-up dynamic controller control parameter information provided by the controller coefficient scheduler and the fourth engine status information provided by the signal processor, and send the obtained calculation result to the fuel control amount selector.

[0019] Furthermore, in some embodiments of the present invention, the rolling optimization prediction closed-loop controller further includes an engine state scheduler. The engine state scheduler is used to determine the engine's current state based on information provided by the signal processor, and to set the engine's operating strategy signal according to the current state, outputting the operating strategy signal to the fuel control quantity selector.

[0020] Furthermore, in some embodiments of the present invention, the fuel control quantity selector selects the fuel control quantity sent by the thrust prediction closed-loop controller, the steady-state controller, the transient controller and the engine dynamic controller according to the working strategy signal from the engine state scheduler, so as to output the corresponding fuel command.

[0021] Furthermore, in some embodiments of the present invention, the operating strategy selection logic of the fuel control quantity selector includes: when the operating strategy is thrust prediction mode, using the fuel command of the fuel control quantity sent by the thrust prediction closed-loop controller, without distinguishing between steady state, transient state, and start-up dynamic state; when the operating strategy is normal speed control mode, calculating the fuel command based on the fuel control quantity sent by the transient controller, the start-up dynamic controller, and the steady state controller, and using a high-low selection rule.

[0022] Furthermore, in some embodiments of the present invention, the rolling optimization prediction closed-loop controller further includes a servo actuator closed-loop controller. The servo actuator closed-loop controller acquires the fuel command output by the fuel control quantity selector and converts it into a desired position signal for the metering mechanism. Simultaneously, it receives the actual position feedback signal of the metering mechanism input by the signal processor, and calculates the servo control current based on the deviation between the two signals.

[0023] Furthermore, the engine rolling optimization predictive closed-loop control system provided according to the second aspect of the present invention includes a rolling optimization predictive closed-loop controller as provided in the first aspect of the present invention, an airborne sensor, and an actuator. The airborne sensor is connected to the rolling optimization predictive closed-loop controller and is used to collect engine body signals and send them to the rolling optimization predictive closed-loop controller. The actuator is connected to the rolling optimization predictive closed-loop controller and is used to execute corresponding actions according to the control commands output by the rolling optimization predictive closed-loop controller. Attached Figure Description

[0024] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0025] Figure 1 A schematic diagram of the framework of an engine rolling optimization predictive closed-loop control system provided according to some embodiments of the present invention is shown.

[0026] Figure 2 A schematic diagram of the framework of a rolling optimization prediction closed-loop controller provided according to some embodiments of the present invention is shown.

[0027] Figure 3 A schematic diagram of the frame of a thrust prediction closed-loop controller provided according to some embodiments of the present invention is shown.

[0028] Figure label: 10 Rolling Optimization Predictive Closed-Loop Controller 111 Thrust Manager 112 Signal Processor 113 Controller Coefficient Scheduler 12 Thrust Prediction Closed-Loop Controller 121 Initial Value Calculation Module 122 Discrete Optimization Matrix Calculation Module 123 Initial Control Law Calculation Module 124 Control Law Calculation Module 125 Sample Value Update Calculation Module 126 Jacobi Matrix Calculation Module 127 Control Law Change Rate Calculation Module 128 Recursive Least Squares Calculation Module 13 Steady-state controller 14 Transient Controller 15 dynamic controllers 16 Fuel Control Quantity Selector 17. Fault Diagnostic Tool 18 Thrust Program Converter 19 Servo Actuator Closed-Loop Controller 20 Airborne Sensors 30. Implementing agencies 40 Engine Rolling Optimization Predictive Closed-Loop Control System 50 Engine body 60 Aircraft hardwire signal 70 Aircraft bus signals Detailed Implementation

[0029] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0030] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0031] Furthermore, the terms "upper," "lower," "left," "right," "top," "bottom," "horizontal," and "vertical" used in the following description should be understood as the orientations shown in the relevant paragraphs and accompanying drawings. These relative terms are for illustrative purposes only and do not imply that the described apparatus must be manufactured or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0032] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0033] As mentioned above, there are very few reports on major engine OEMs (Original Equipment Manufacturers) regarding their plans to directly control engine thrust using an onboard controller in a closed-loop manner. In engineering practice, the relationship between thrust and engine speed is typically calibrated on a test bench (ground, in flight, etc.), and a converted tachometer and thrust curve are used to control the converted speed to achieve thrust control. This approach requires calibration under various engine operating scenarios, has high development costs, and exhibits poor dynamic adaptive capabilities.

[0034] Furthermore, Chinese invention CN201710931188.5 proposes an energy management method for hybrid-powered ships based on model predictive control (MMC). It selects a Markov model to predict the power demand of the hybrid-powered ship and, combined with the principles of MMC, uses dynamic programming to optimize power allocation within the prediction time domain with the objective of minimizing fuel consumption. However, this method suffers from the problem of cumbersome theoretical solution steps. Chinese invention CN201711134813.X discloses an algorithm for estimating the thrust of aero-engines using an adaptive RBF neural network, proposing an adaptive RBF neural network based on particle swarm optimization (PSO). In the improved PSO algorithm, a local optimum is set with the same number of hidden layer node types as the number of hidden layer node types. This method also suffers from the problem of cumbersome theoretical solution steps.

[0035] To overcome the aforementioned shortcomings of existing technologies, this invention provides a rolling optimization predictive closed-loop controller and an engine rolling optimization predictive closed-loop control system. This controller can achieve the functions of traditional steady-state, transient, and dynamic controllers, thereby realizing closed-loop control for dynamic engine thrust prediction and simplifying the controller architecture. Furthermore, the fuel control selector utilizes a neural network model to calculate fuel control quantities, simplifying the solution process for dynamic engine thrust predictive control and improving thrust prediction performance, thus enhancing the safety margin of engine thrust control.

[0036] In some non-limiting embodiments, the rolling optimization prediction closed-loop controller provided in the first aspect of the present invention can be configured to be implemented in the engine rolling optimization prediction closed-loop control system provided in the second aspect of the present invention.

[0037] Please refer to the details. Figure 1 . Figure 1 A schematic diagram of the framework of an engine rolling optimization predictive closed-loop control system provided according to some embodiments of the present invention is shown.

[0038] exist Figure 1In the illustrated embodiment, the engine rolling optimization predictive closed-loop control system provided by the second aspect of the present invention includes a rolling optimization predictive closed-loop controller 10, an airborne sensor 20 for the engine, and an actuator 30, all provided by the first aspect of the present invention. Here, the rolling optimization predictive closed-loop controller 10 is used to acquire and receive aircraft hardwired signals 60 and aircraft bus signals 70, and is suitable for online, real-time control. It can obtain the engine dynamic model through offline theoretical calculations or experiments, ensuring that the engine operates along the desired thrust curve while guaranteeing optimal performance indicators. The airborne sensor 20 is connected to the rolling optimization predictive closed-loop controller 10 and is used to acquire engine body signals 50 and send them to the rolling optimization predictive closed-loop controller 10. The actuator 30 is connected to the rolling optimization predictive closed-loop controller 10 and is used to execute corresponding actions according to the control commands output by the rolling optimization predictive closed-loop controller 10.

[0039] Please refer to further details. Figure 2 . Figure 2 A schematic diagram of the framework of a rolling optimization prediction closed-loop controller provided according to some embodiments of the present invention is shown.

[0040] exist Figure 2 In the illustrated embodiment, the first aspect of the present invention provides the above-mentioned rolling optimization prediction closed-loop controller including an engine information calculation module, a thrust prediction closed-loop controller 12, a steady-state controller 13, a transient controller 14, a dynamic controller 15, and a fuel control quantity selector 16.

[0041] Specifically, this engine information calculation module is used to determine the required engine dynamic model information, as well as the engine state information corresponding to the engine throttle lever angle signal PLA and the thrust mode signal ThrustMode, based on the engine's throttle lever angle signal PLA and thrust mode signal ThrustMode. Here, the engine dynamic information includes at least the engine's temperature and pressure. The unit of the throttle lever angle signal PLA is degrees.

[0042] The thrust prediction closed-loop controller 12 is used to calculate the fuel control quantity Wf_MPC required for the thrust prediction closed loop based on the engine dynamic model information and the first engine state information provided by the engine information calculation module, and sends the obtained first calculation result Wf_MPC signal to the fuel control quantity selector 16. Here, the unit of fuel control quantity Wf_MPC is kg / h.

[0043] The steady-state controller 13 calculates the engine steady-state fuel control quantity Wf_Steady required for the speed closed loop based on the engine dynamic model information and the second engine state information provided by the engine information calculation module, and sends the obtained second calculation result Wf_Steady signal to the fuel control quantity selector 16. Here, the unit of fuel control quantity Wf_Steady is kg / h.

[0044] The transient controller 14 calculates the required engine acceleration fuel control quantity Wf_Acc and engine deceleration fuel control quantity Wf_Dec based on the engine dynamic model information and third engine state information provided by the engine information calculation module, and sends the obtained third calculation results Wf_Acc and Wf_Dec signals to the fuel control quantity selector 16. Here, the units of fuel control quantities Wf_Acc and Wf_Dec are kg / h.

[0045] The dynamic controller 15 is used to calculate the engine ignition fuel control quantity Wf_Start required for starting control based on the engine dynamic model information and the fourth engine state information provided by the engine information calculation module, and sends the obtained fourth calculation result Wf_Start signal to the fuel control quantity selector 16.

[0046] The fuel control quantity selector 16 is used to input the first calculation result, the second calculation result, the third calculation result and / or the fourth calculation result into a pre-trained neural network model to output the corresponding fuel control quantity, thereby outputting the corresponding fuel command.

[0047] Specifically, during the training of the neural network model, the rolling optimization predictive closed-loop controller 10 can acquire historical operating data of the engine. This historical operating data includes historical engine state information and corresponding fuel control quantities. Then, the rolling optimization predictive closed-loop controller 10 can construct the neural network model to be trained. Finally, the rolling optimization predictive closed-loop controller 10 can train the neural network model based on the historical operating data.

[0048] Furthermore, in Figure 1 In the illustrated embodiment, the engine information calculation module provided by the first aspect of the present invention includes a thrust manager 111, a signal processor 112, and a controller coefficient scheduler 113. Here, the thrust manager 111 calculates the engine operating thrust expectation command Fd based on signals such as the engine throttle lever angle signal PLA and the thrust mode signal ThrustMode, and outputs it to the thrust prediction closed-loop controller 12.

[0049] The signal processor 112 calculates the true physical value of the converted signal based on the acquired original sensor signal and outputs it to the thrust prediction closed-loop controller 12, steady-state controller 13, transient controller 14 and dynamic controller 15.

[0050] The controller coefficient scheduler 113 receives the turbofan engine high-pressure rotor converted speed signal N2r25 sent by the signal processor 112, and combines it with the stored offline acquired information to determine the engine dynamic model information required by the thrust prediction closed-loop controller 12.

[0051] Specifically, the control parameters provided by the controller coefficient scheduler 113 to the steady-state controller 13 include the conventional controller parameters Kpstd and Kird. The control parameters provided by the controller coefficient scheduler 113 to the transient controller 14 and the dynamic controller 15 include the predictive controller parameters Kptr and Kitr. The controller coefficient scheduler 113 provides the required engine dynamic model information to the thrust prediction closed-loop controller 12, which is the engine dynamic model function. The expression coefficients.

[0052] Furthermore, in some embodiments, the engine dynamic model is as follows:

[0053] in, The dynamic equations for the engine are as follows. The motion state of the dynamic equation. For the input quantities of the dynamic equation, For the engine of the Loyabi, The coefficients are linear. This is the gain coefficient. This is the static pressure at the outlet of the high-pressure compressor, expressed in kPa. The engine's operating ambient temperature. This represents the expected thrust of the engine during operation. This represents the actual thrust generated by the engine, measured in kN. This refers to the fuel control quantity, expressed in kg / h.

[0054] In addition, Figure 1In the illustrated embodiment, the rolling optimization prediction closed-loop controller 10 provided in the first aspect of the present invention further includes a fault diagnostic unit 17. This fault diagnostic unit 17 is used to receive abnormal signals from the engine information calculation module 11, the thrust prediction closed-loop controller 12, the steady-state controller 13, the transient controller 14, the dynamic controller 15, and / or the fuel control quantity selector 16, for signal anomaly diagnosis. Alternatively, the fault diagnostic unit 17 is used to detect the engine's hardware status for hardware anomaly diagnosis.

[0055] Please refer to Figure 3 . Figure 3 A schematic diagram of the frame of a thrust prediction closed-loop controller provided according to some embodiments of the present invention is shown.

[0056] exist Figure 3 In the embodiment shown, the thrust prediction closed-loop controller 12 includes an initial value calculation module 121, a discrete optimization matrix calculation module 122, an initial control law calculation module 123, a control law calculation module 124, a sampled value update calculation module 125, a Jacobi matrix calculation module 126, a control law change rate calculation module 127, and a recursive least squares calculation module 128.

[0057] Here, the initial value calculation module 121 is connected to the discrete optimization matrix calculation module 122 and the initial control law calculation module 123 to perform initial value calculation and calculate the initial values ​​of each variable in the thrust prediction closed-loop controller 12.

[0058] Furthermore, the variables in the thrust prediction closed-loop controller 12 include: a) t=0, where t is time in seconds; b) △T = Ts, where Ts is the sampling time in seconds; c) X(k) = X_sample(k△T), where X_sample(k△T) is the sampled state value at time k△T, k = 0, 1, 2...; d) X0=X(0), where X0 is the initial value variable and X(0) is the value at the first position of the state vector X; e) t =0, where, t For U t The rate of change of U over time t To augment the control matrix; f) A, b, where A is [1, 0; 0, 1] and b is [1; 1]; g) firstflag1=0, where firstflag1 is the first intermediate variable; h) firstflag2=0, where firstflag2 is the second intermediate variable.

[0059] In addition, the discrete optimization matrix calculation module 122 is connected to the initial control law calculation module 123 and is used to perform discrete optimization matrix calculation, and complete the calculation of the controller optimal matrix F according to the following formula.

[0060] Specifically, the discrete optimization matrix calculation module 122 can calculate the C matrix and convert inequalities into equations:

[0061] Here, the symbol ":=" defines the action, that is... Defined as a subsequent formula.

[0062]

[0063] Here, the superscript T indicates transpose.

[0064] Where X represents the motion state of the dynamic equation. For the defined augmented control quantity, u For the input quantities of the dynamic equation, for The third element, for The fourth element, for The fifth element, The coefficients are linear. The minimum value of the linear coefficient. For fuel control quantity, This represents the maximum value of the fuel control quantity. This is the minimum fuel control quantity.

[0065] The discrete optimization matrix calculation module 122 can also calculate the H matrix:

[0066] Where X represents the motion state of the dynamic equation. For the defined augmented control quantity, , For calculation process parameters, For the engine dynamic equations, , and For calculation process parameters, For the expected thrust of the engine, The actual thrust generated by the engine during operation. The coefficients are linear. This is the gain coefficient. t For a moment, N To augment the control matrix Dimensions.

[0067] The discrete optimization matrix calculation module 122 can also calculate the controller optimal matrix F using the C matrix and the H matrix:

[0068] in, For the defined augmented control quantity, , For calculation process parameters, X The motion state of the dynamic equation. N for Dimensions.

[0069] Furthermore, the initial control law calculation module 123 calculates the approximate solution of the initial control law in the least squares sense based on the initial input parameters provided by the initial value calculation module 121, and the H matrix and C matrix provided by the discrete optimal matrix calculation module 122. The initial control law calculation module 123 can calculate the initial control law based on the input from the initial value calculation module 121. and The initial control law is The H matrix and C matrix input by the discrete optimal matrix calculation module 122 are used to calculate the approximate solution μ0 of the following equation in the least squares sense based on the recursive least squares calculation module 128.

[0070]

[0071] in, As the preset initial value of the state variable, take the input. This is the preset initial value of the control quantity.

[0072] Then, the initial control law calculation module 123 outputs the obtained calculation results to the control law calculation module 124.

[0073] In addition, the control law calculation module 124 is connected to the sampled value update calculation module 125, the control law change rate calculation module 127, and the fuel control quantity selector 16. Based on the initial control law and the control law change rate, the fuel control quantity is calculated for each unit of time. The obtained calculation results are output to the sampled value update calculation module 125, and the final fuel control quantity result is sent to the fuel control quantity selector 16.

[0074] The following pseudocode calculation formula is the method for calculating the control quantity for each beat. Firstflag1=1 is the initial time, using the calculation of the first branch, and the calculation of the second branch is used for other times: If (firstflag1 = 1) { Wf (0) = U0[0]; Wf = Wf (0);τ [0,△t) Firstflag1 ++;}

[0075] Else

[0076] { U t+△t =U t + t △t

[0077] Wf t (0)=U t [0];

[0078] Wf(τ) = Wf (0);τ [t, t+△t)

[0079] t = t + Δt;

[0080] Firstflag1 ++;}

[0081] End

[0082] Here, the input parameter of this pseudocode is the initial value of the augmented control variable U. , rate of change over time The output parameter is the initial fuel control quantity Wf. (0) and at time 1 τ The fuel control quantity Wf(τ).

[0083] In addition, the sampled value update calculation module 125 is connected to the Jacobian matrix calculation module 126. According to the calculation results provided by the control law calculation module 124, the sampled value of the fuel control quantity per unit time is updated and calculated, and the obtained calculation results are output to the Jacobian matrix calculation module 126.

[0084] Specifically, the sampled value update calculation module 125 updates the value according to U. t rate of change over time t Then, by integrating over a time step Δt, based on the current U... t Find the value U at the next sampling time. t+△t : X t =X(t +△t); X t + △T = X(t + Δt); = (X t + △t - X t ) / △t.

[0085] Here, t is the sampled value calculated at the current time, and t+△t is the value calculated at the next sampling time.

[0086] In addition, the Jacobian matrix calculation module 126 performs update calculation of the Jacobian matrix per unit time for each cycle based on the initial input parameters provided by the initial value calculation module 121, and outputs the obtained calculation results to the control law change rate calculation module 127.

[0087] The following pseudocode calculation formula is the method for updating the Jacobian matrix at each time step. Firstflag2=1 is the initial time step, which uses the calculation of the first branch. Other time steps use the calculation of the second branch: If (firstflag2=1) { t_est = 0;}

[0088] Else

[0089] { t_est = t-△t ;}

[0090] End

[0091] b: = -ζ F(U t ,X t ,t) – (F(U t ,X t +△t t , t+h) - F(U t ,X t , t)) / △t;

[0092] A: = -(F(U t +△t t_est ,X t +△t t , t+h) - F(U t ,X t +△t t , t+△t)) / △t;

[0093] in, t_est At time t t The estimated values ​​are given by ζ, b, and A, which are parameters of the Jacobian matrix.

[0094] Furthermore, the control law rate of change calculation module 127 calculates the approximate solution of the control law rate of change in the least squares sense based on the initial input parameters provided by the initial value calculation module and the recursive least squares calculation module 128, and then outputs the obtained calculation result to the control law calculation module 127. Here, the control law rate of change is... t Given A and b input from the initial value calculation module 151, the control law rate of change calculation module 127 can, based on the recursive least squares calculation module 158, complete the following equation as an approximate solution in the least squares sense. t calculate: b - A t = 0 in, t For U t The rate of change with respect to time, b and A are the parameters of the Jacobian matrix.

[0095] In addition, the recursive least squares calculation module 128 performs least squares meaning equation solving calculations based on the rotational transformation method for the input equation.

[0096] In addition, Figure 1In the illustrated embodiment, the rolling optimization prediction closed-loop controller of the present invention further includes a thrust planning converter 18. Here, the thrust planning converter 18 is connected to the thrust manager 111 and is used to perform target calculations for steady-state control based on the engine operating thrust expectation command Fd sent by the thrust manager 111, so as to obtain the first target control speed N1Cmd of the low-pressure shaft of the turbofan engine and the second target control speed N2Cmd of the high-pressure shaft of the turbofan engine, and output them to the steady-state controller 13.

[0097] Furthermore, the controller coefficient scheduler 113 provides the steady-state controller 13 with the necessary control parameters and the calculation output function of existing patterns. The steady-state controller 13 is connected to the thrust planning converter 18 and is used to calculate the engine steady-state fuel control quantity required for the speed closed loop based on the steady-state controller 13 control parameter information provided by the controller coefficient scheduler 113 and the second engine state information provided by the signal processor 112, and sends the obtained calculation results to the fuel control quantity selector. Here, the second engine state information provided by the signal processor 112 includes the turbofan engine low-pressure shaft speed N1 and the turbofan engine high-pressure shaft speed N2, with N1 and N2 in rpm.

[0098] Furthermore, the calculation method for the engine steady-state fuel control quantity Wf_Steady can employ traditional proportional-derivative-integral controllers, sliding mode controllers, and state feedback mechanisms.

[0099] Alternatively, the thrust planning converter 18 performs target calculations for transient control based on the engine operating thrust expectation command Fd sent by the thrust manager 111, to obtain the rate of change of the first target control speed N1dotCmd of the low-pressure shaft of the turbofan engine and the rate of change of the second target control speed N2dotCmd of the high-pressure shaft of the turbofan engine, and outputs them to the transient controller 14.

[0100] Furthermore, the controller coefficient scheduler 113 provides the transient controller 14 with the necessary control parameters and the calculation output function of existing patterns. The transient controller 14 is connected to the thrust planning converter 18 and is used to calculate the engine acceleration fuel control quantity and engine deceleration fuel control quantity required for acceleration control based on the control parameter information of the transient controller 14 provided by the controller coefficient scheduler 113 and the third engine state information provided by the signal processor 112, and sends the obtained calculation results to the fuel control quantity selector 16. Here, the third engine state information provided by the signal processor 112 includes the turbofan engine high-pressure rotor converted speed N2r25 and the rate of change of the turbofan engine high-pressure shaft speed N2dot, where N2r25 is in rpm and N2dot is in rpm / s.

[0101] Furthermore, the calculation methods for engine acceleration fuel control quantity Wf_Acc and engine deceleration fuel control quantity Wf_Dec can employ traditional proportional-derivative-integral controllers, sliding mode controllers, and feedforward series integral controllers.

[0102] Alternatively, the thrust planning converter 18 performs target calculations for dynamic control based on the engine operating thrust expectation command Fd sent by the thrust manager 111, to obtain the rate of change of the first target control speed N1dotCmd of the low-pressure shaft of the turbofan engine and the rate of change of the second target control speed N2dotCmd of the high-pressure shaft of the turbofan engine, and outputs them to the dynamic controller 15.

[0103] Furthermore, the controller coefficient scheduler 113 provides the required control parameters and existing calculation output functions to the start-up dynamic controller 15. The start-up dynamic controller 15 is connected to the thrust planning converter 18 and is used to calculate the engine starting fuel control quantity required for starting control based on the control parameter information provided by the controller coefficient scheduler 113 and the fourth engine status information provided by the signal processor 112, and sends the obtained calculation results to the fuel control quantity selector 16. Here, the fourth engine status information provided by the signal processor 112 includes the turbofan engine high-pressure rotor converted speed N2r25 and the rate of change of the turbofan engine high-pressure shaft speed N2dot, where N2r25 is in rpm and N2dot is in rpm / s.

[0104] Furthermore, in some embodiments, the rolling optimization prediction closed-loop controller provided in the first aspect of the present invention further includes an engine state scheduler, which is used to determine the state of the engine based on information provided by the signal processor, and to set the engine working strategy signal according to the current state, and output the working strategy signal to the fuel control quantity selector 16.

[0105] Furthermore, the fuel control quantity selector 16 selects the fuel control quantity sent by the thrust prediction closed-loop controller 12, steady-state controller 13, transient controller 14 and dynamic controller 15 according to the working strategy signal from the engine state scheduler, so as to output the corresponding fuel command.

[0106] Furthermore, the operating strategy selection logic of the fuel control quantity selector 16 is as follows: when the operating strategy is thrust prediction mode, it uses the fuel command of the fuel control quantity sent by the thrust prediction closed-loop controller, without distinguishing between steady state, transient state, and start-up dynamic state. When the operating strategy is normal speed control mode, it calculates the fuel command based on the fuel control quantities sent by the transient controller, start-up dynamic controller, and steady state controller, and adopts a high-low selection rule.

[0107] In addition, Figure 1In the embodiment shown, the rolling optimization prediction closed-loop controller provided by the first aspect of the present invention further includes a servo actuator closed-loop controller 19, which acquires the fuel command output by the fuel control quantity selector 16 and converts it into the position expectation signal of the metering mechanism, while receiving the actual position feedback signal of the metering mechanism input by the signal processor, so as to calculate the servo control current based on the deviation between the two.

[0108] In summary, the rolling optimization predictive closed-loop controller and engine rolling optimization predictive closed-loop control system provided by this invention can both utilize the thrust prediction closed-loop controller to achieve the functions of traditional steady-state, transient, and dynamic controllers, thereby realizing closed-loop control of engine dynamic thrust prediction and simplifying the controller form. Furthermore, the fuel control selector uses a neural network model to calculate the fuel control quantity, simplifying the solution process for engine dynamic thrust prediction control, improving thrust prediction performance, and enhancing the safety margin of engine thrust control.

[0109] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0110] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A rolling optimization predictive closed loop controller characterized by, include: An engine information calculation module is used to determine the required engine dynamic model information and engine state information corresponding to the engine throttle lever angle signal and the thrust mode signal based on the engine throttle lever angle signal and the thrust mode signal, wherein the engine dynamic information includes at least the engine temperature and pressure. The thrust prediction closed-loop controller is used to calculate the fuel control amount required for the thrust prediction closed loop based on the engine dynamic model information and the first engine state information provided by the engine information calculation module, and send the obtained first calculation result to the fuel control amount selector. The steady-state controller is used to calculate the engine steady-state fuel control quantity required for the speed closed loop based on the engine dynamic model information and the second engine state information provided by the engine information calculation module, and to send the obtained second calculation result to the fuel control quantity selector. The transitional controller is used to calculate the engine acceleration fuel control amount and engine deceleration fuel control amount required for acceleration control based on the engine dynamic model information and third engine state information provided by the engine information calculation module, and send the obtained third calculation result to the fuel control amount selector. The dynamic controller is used to calculate the engine ignition fuel control quantity required for starting control based on the engine dynamic model information and fourth engine state information provided by the engine information calculation module, and to send the obtained fourth calculation result to the fuel control quantity selector; and The fuel control quantity selector is used to input the first calculation result, the second calculation result, the third calculation result and / or the fourth calculation result into a pre-trained neural network model to output the corresponding fuel control quantity, thereby outputting the corresponding fuel command.

2. The roll optimization predictive closed loop controller of claim 1 wherein, The rolling optimization prediction closed-loop controller also includes a fault diagnostic unit, wherein... The fault diagnostic device is used to receive abnormal signals from the engine information calculation module, the thrust prediction closed-loop controller, the steady-state controller, the transient controller, the engine dynamic controller, and / or the fuel control quantity selector, to diagnose these signal abnormalities; and / or The fault diagnostic tool is used to detect the hardware status of the engine in order to diagnose hardware anomalies.

3. The roll optimization predictive closed loop controller of claim 1 wherein, The steps for training the neural network model include: Acquire historical operating data of the engine, wherein the historical operating data includes historical state information of the engine and corresponding fuel control quantity; Constructing a neural network model to be trained; and The neural network model is trained based on the historical operating data.

4. The roll optimization predictive closed loop controller of claim 1 wherein, The engine information calculation module includes: The thrust manager calculates the expected operating thrust command of the engine based on the throttle lever angle signal and thrust mode signal of the engine, and outputs it to the thrust prediction closed-loop controller. The signal processor calculates the true physical value of the converted signal based on the acquired raw sensor signal and outputs it to the thrust prediction closed-loop controller, the steady-state controller, the transient controller and the dynamic controller. The controller coefficient scheduler receives the converted speed signal of the turbofan engine high-pressure rotor sent by the signal processor, and determines the engine dynamic model information required by the thrust prediction closed-loop controller by combining it with the stored offline acquired information.

5. The rolling optimization predictive closed-loop controller as described in claim 4, characterized in that, The engine dynamic model is shown below: in, Here is the dynamic equation for the engine. The motion state of the dynamic equation. represents the input quantity for the dynamic equation. The loyabi of the engine, The coefficients are linear. This is the gain coefficient. This refers to the static pressure at the outlet of the high-pressure compressor. The operating ambient temperature of the engine. The expected operating thrust of the engine. The actual thrust generated by the engine during operation. This refers to the fuel control quantity.

6. The rolling optimization predictive closed-loop controller as described in claim 5, characterized in that, The thrust prediction closed-loop controller includes an initial value calculation module, a discrete optimization matrix calculation module, an initial control law calculation module, a control law calculation module, a sampled value update calculation module, a Jacobian matrix calculation module, a control law change rate calculation module, and a recursive least squares calculation module. The initial value calculation module is used to preset the initial input parameters for the corresponding variables. The discrete optimization matrix calculation module is used to calculate the optimal C matrix and H matrix of the controller. The initial control law calculation module calculates the approximate solution of the initial control law in the least squares sense based on the initial input parameters provided by the initial value calculation module, and the H and C matrices provided by the discrete optimization matrix calculation module, and then outputs the obtained calculation results to the control law calculation module. The control law calculation module calculates the fuel control quantity per unit time for each cycle based on the initial control law and the rate of change of the control law. The calculation result is output to the sampled value update calculation module, and the final fuel control quantity result is sent to the fuel control quantity selector. The sampled value update calculation module updates the sampled value of the fuel control quantity for each unit of time according to the calculation results provided by the control law calculation module, and outputs the obtained calculation results to the Jacobian matrix calculation module. The Jacobian matrix calculation module updates the Jacobian matrix for each unit of time step based on the initial input parameters provided by the initial value calculation module, and outputs the obtained calculation results to the control law change rate calculation module. The control law rate of change calculation module calculates the approximate solution of the control law rate of change in the least squares sense based on the initial input parameters provided by the initial value calculation module and the recursive least squares calculation module, and then outputs the obtained calculation result to the control law calculation module. The recursive least squares calculation module solves the least squares meaning equation of the input equation based on the rotational transformation method.

7. The rolling optimization predictive closed-loop controller as described in claim 6, characterized in that, The rolling optimization prediction closed-loop controller also includes a thrust planning converter, wherein... The thrust planning converter is connected to the thrust manager and is used to perform steady-state control target calculations based on the engine operating thrust expectation command sent by the thrust manager, to obtain the first target control speed of the low-pressure shaft of the turbofan engine and the second target control speed of the high-pressure shaft of the turbofan engine, and output them to the steady-state controller; and / or The thrust planning converter performs target calculations for transient control based on the engine operating thrust expectation command sent by the thrust manager, to obtain the rate of change of the first target control speed of the turbofan engine's low-pressure shaft and the rate of change of the second target control speed of the turbofan engine's high-pressure shaft, and outputs them to the transient controller; and / or The thrust planning converter performs target calculations for dynamic control based on the engine operating thrust expectation command sent by the thrust manager, so as to obtain the rate of change of the first target control speed of the low-pressure shaft of the turbofan engine and the rate of change of the second target control speed of the high-pressure shaft of the turbofan engine, and outputs them to the dynamic controller.

8. The rolling optimization predictive closed-loop controller as described in claim 7, characterized in that, The controller coefficient scheduler is used to provide the steady-state controller with the required control parameters and the calculation output function of existing patterns. The steady-state controller is connected to the thrust planning converter and is used to calculate the engine steady-state fuel control quantity required for speed closed-loop based on the steady-state controller control parameter information provided by the controller coefficient scheduler and the second engine state information provided by the signal processor, and send the obtained calculation result to the fuel control quantity selector.

9. The rolling optimization predictive closed-loop controller as described in claim 7, characterized in that, The controller coefficient scheduler is used to provide the transient controller with the required control parameters and the calculation output function of existing rules. The transient controller is connected to the thrust planning converter and is used to calculate the engine acceleration fuel control amount and engine deceleration fuel control amount required for acceleration control based on the transient controller control parameter information provided by the controller coefficient scheduler and the third engine state information provided by the signal processor, and send the obtained calculation results to the fuel control amount selector.

10. The rolling optimization predictive closed-loop controller as described in claim 7, characterized in that, The controller coefficient scheduler is used to provide the dynamic controller with the required control parameters and the calculation output function of existing rules. The start dynamic controller is connected to the thrust planning converter and is used to calculate the engine start fuel control amount required for start control based on the start dynamic controller control parameter information provided by the controller coefficient scheduler and the fourth engine status information provided by the signal processor, and send the obtained calculation result to the fuel control amount selector.

11. The rolling optimization predictive closed-loop controller as described in claim 7, characterized in that, Also includes: The engine state scheduler is used to determine the state of the engine based on the information provided by the signal processor, and to set the engine working strategy signal according to the current state, and output the working strategy signal to the fuel control quantity selector.

12. The rolling optimization predictive closed-loop controller as described in claim 11, characterized in that, The fuel control quantity selector selects the fuel control quantity sent by the thrust prediction closed-loop controller, the steady-state controller, the transient controller, and the engine dynamic controller based on the working strategy signal from the engine state scheduler, so as to output the corresponding fuel command.

13. The rolling optimization predictive closed-loop controller as described in claim 12, characterized in that, The operating strategy selection logic of the fuel control quantity selector includes: When the working strategy is thrust prediction mode, the fuel command for fuel control quantity sent by the thrust prediction closed-loop controller is used, without distinguishing between steady state, transient state and start-up dynamic state. When the operating strategy is in normal speed control mode, the fuel command is calculated based on the fuel control quantities sent by the transient controller, the starting dynamic controller and the steady-state controller, and using a high-low selection rule.

14. The rolling optimization predictive closed-loop controller as described in claim 13, characterized in that, Also includes: The servo actuator closed-loop controller acquires the fuel command output by the fuel control quantity selector and converts it into the desired position signal of the metering mechanism. At the same time, it receives the actual position feedback signal of the metering mechanism input by the signal processor, so as to calculate the servo control current based on the deviation between the two.

15. An engine rolling optimization predictive closed-loop control system, comprising a rolling optimization predictive closed-loop controller as described in any one of claims 1 to 14, an airborne sensor, and an actuator, wherein, The airborne sensor is connected to the rolling optimization prediction closed-loop controller and is used to collect engine body signals and send them to the rolling optimization prediction closed-loop controller. The actuator is connected to the rolling optimization prediction closed-loop controller and is used to execute corresponding actions according to the control commands output by the rolling optimization prediction closed-loop controller.

Images