CN-116841202-B - Nested optimization method and device for accelerating control plan of aero-engine

CN116841202BCN 116841202 BCN116841202 BCN 116841202BCN-116841202-B

Abstract

The invention discloses an aircraft engine acceleration control plan nesting optimization method, and belongs to the technical field of aircraft engine control. The method uses an inner loop and outer loop nesting mode to conduct nesting optimization on geometric mechanisms participating in aeroengine control, wherein the geometric mechanisms participating in open loop control are globally optimized in an outer loop through construction of Bezier curves, the geometric mechanisms participating in closed loop control are locally roll optimized in an inner loop, the optimization of the inner loop is based on the outer loop, and the optimization of the outer loop uses acceleration response time of the inner loop as an evaluation index. The invention also discloses a nested optimizing device for the accelerating control plan of the aeroengine. Compared with the prior art, the method has the advantages that the control plan optimization process is more consistent with the actual control process, the advantages of multi-variable control can be fully exerted, and the accelerated evaluation process is more direct.

Inventors

LI QIUHONG
ZHAO XINGYU
LIU XINYANG
GU ZIYU
PANG SHUWEI

Assignees

南京航空航天大学

Dates

Publication Date: 20260505
Application Date: 20230703

Claims (8)

1. A nested optimization method for an accelerating control plan of an aeroengine is characterized by comprising the steps of performing nested optimization on geometric mechanisms participating in control of the aeroengine in an inner loop and outer loop nested mode, performing global optimization on the geometric mechanisms participating in open loop control by constructing a Bezier curve in the outer loop, performing local rolling optimization on the geometric mechanisms participating in closed loop control in the inner loop, wherein the optimization of the inner loop is based on the outer loop, the optimization of the outer loop takes the accelerating response time of the inner loop as an evaluation index, the inner loop is used for online construction of a prediction model based on a neural network state space model and performing local rolling optimization on closed loop control variables by using an alternating direction multiplier method, and the neural network state space model comprises an aeroengine state space model and a neural network model, wherein parameters of the aeroengine state space model are described by the neural network model parameters and updated along with online updating of the neural network model parameters.
2. The nested optimization method of an aeroengine acceleration control plan according to claim 1, wherein the global optimization is performed in an outer loop by using a group intelligent optimization method, individual dimensions are determined according to the number of geometric mechanisms participating in open loop control and Bezier curve control points, and a generation range of an initial population is determined according to an adjustment range of the geometric mechanisms.
3. The method of nested optimization of an aircraft engine acceleration control scheme of claim 1, wherein the control points of the bezier curve consist of rotational speed and geometry positions involved in open loop control.
4. The method for nested optimization of an aircraft engine acceleration control scheme according to claim 1, wherein the neural network comprises an hidden layer, an output layer and a multiplication layer arranged between the hidden layer and the output layer, the multiplication layer takes state quantity and control quantity of an aircraft engine state space model as an excitation function, the hidden layer is divided into n+p groups according to the dimension of the state quantity and the control quantity, n represents the dimension of a state variable, p represents the dimension of an input variable, each group comprises j hidden layer nodes with the same number, the output of each group of hidden layers is multiplied by the state variable x and the input variable u at the multiplication layer, and the connection weight between the multiplication layer and the output layer is calculated according to a recursive least square method.
5. The nested optimizing device for the accelerating control plan of the aeroengine is characterized by comprising an inner loop and an outer loop which are nested with each other and used for conducting nested optimization on geometric mechanisms participating in control of the aeroengine, wherein the outer loop is used for conducting global optimization on the geometric mechanisms participating in open loop control by constructing a Bezier curve, the inner loop is used for conducting local rolling optimization on the geometric mechanisms participating in closed loop control, the optimization of the inner loop is based on the outer loop, the optimization of the outer loop is based on the accelerating response time of the inner loop as an evaluation index, a prediction model is built on line on the basis of a neural network state space model and conducting local rolling optimization on closed loop control variables by using an alternating direction multiplier method, the neural network state space model comprises an aeroengine state space model and a neural network model, and parameters of the aeroengine state space model are described by the neural network model parameters and updated along with on-line updating of the neural network model parameters.
6. The nested optimizing device for the accelerating control plan of the aeroengine according to claim 5, wherein the outer loop performs global optimization by using a group intelligent optimizing method, the individual dimension is determined according to the number of geometric mechanisms participating in open loop control and Bezier curve control points, and the generation range of an initial population is determined according to the adjustment range of the geometric mechanisms.
7. The nested optimizing device for an aircraft engine acceleration control scheme of claim 5, wherein the control points of the bezier curve are comprised of rotational speed and geometry positions involved in open loop control.
8. The nested optimizing device for the acceleration control plan of the aeroengine according to claim 5, wherein the neural network comprises an hidden layer, an output layer and a multiplication layer arranged between the hidden layer and the output layer, the multiplication layer takes a state quantity and a control quantity of a state space model of the aeroengine as an excitation function, the hidden layer is divided into n+p groups according to the dimension of the state quantity and the control quantity, n represents the dimension of a state variable, p represents the dimension of an input variable, each group comprises j hidden layer nodes with the same number, the output of each group of hidden layers is multiplied by the state variable x and the input variable u at the multiplication layer, and the connection weight between the multiplication layer and the output layer is calculated according to a recursive least square method.

Description

Nested optimization method and device for accelerating control plan of aero-engine Technical Field The invention relates to the technical field of aero-engine control, in particular to an aero-engine acceleration control plan optimization method. Background An important operation of the engine is an acceleration/deceleration operation. In order to ensure that the engine does not overheat, surge or stall in the acceleration and deceleration process and can quickly respond, it is important to design an optimal acceleration and deceleration control rule, and in the design of the acceleration and deceleration control rule, the acceleration and deceleration control plan directly influences the response performance and the safety of the aeroengine, so that the optimization technology is widely adopted to optimize the acceleration and deceleration control plan. Common acceleration and deceleration (hereinafter collectively referred to as acceleration) control plan optimization methods include methods based on modern artificial intelligence and methods based on classical nonlinear optimization. For example, in a study on turbofan engines, literature [1] combines a genetic algorithm and a sequential quadratic programming algorithm, and performs optimization solving on an acceleration process in which engine thrust reaches maximum in a shortest time. The literature [2] establishes an optimization problem for the whole acceleration process of the turbofan engine, optimizes Bezier curves of main fuel flow and tail nozzle throat area by adopting a sequential secondary planning method, and further constructs an acceleration control plan. The document [3] [4] expands the optimization of the control plan to the full envelope, and establishes a converted gas-oil ratio control plan based on an equal temperature line and an N-dot control plan based on an equal altitude line. And the documents [5] and [6] develop transition state control plan designs of the variable cycle engine by means of a multidisciplinary optimization design ISIGHT platform and through the maximization of residual power. The control plan optimization processes are not used for distinguishing the open-loop control geometric mechanism from the closed-loop control geometric mechanism, the same optimization method is adopted, or only the closed-loop control geometric mechanism is optimized, so that the advantages of the multivariable control are not exerted. Reference is made to: [1] Shi Peiyan, gou Lin Feng, guo Jiangwei, etc. A GA-SQP based aero engine acceleration optimizing control [ J ]. Computer and modernization 2014,0 (01): 62-66. [2]ZHENG Qiangang,ZHANG Haibo.A global optimization control for turbo-fan engine acceleration schedule design[J].Proceedings of the Institution of Mechanical Engineers,Part G:Journal of Aerospace Engineering,2018,232(2):308-316. [3] Liu Zihe, zhengzheng Steel, liu Minglei, et al turbofan engine full envelope acceleration control plan improvement method research [ J ]. Propulsion techniques, 2022,43 (01): 346-353. [4] Li Yuchen, li Qiugong, zhang Xin et al A turbofan engine N-dot control method based on active switching logic [ J/OL ]. Peking university of aviation, university of Beijing, 1-14[2023-06-28]. Https:// doi.org/10.13700/j.bh.1001-5965.2022.0022. [5] Zhu Baibin design of control plan and control algorithm of three-way engine, nanjing, university of aviation aerospace, nanjing, 2020. [6]JIA L,CHEN Y,CHENG R,et al.Designing method of acceleration and deceleration control schedule for variable cycle engine[J].Chinese Journal of Aeronautics,2021,34(05):27-38. Disclosure of Invention The invention aims to solve the technical problems of overcoming the defects of the prior art and providing an aeroengine acceleration control plan nesting optimization method, wherein an inner loop optimizes the position of a closed-loop geometric mechanism to achieve the optimal control effect, and an outer loop optimizes the position of an open-loop geometric mechanism to improve the acceleration performance of an engine. The technical scheme adopted by the invention is as follows: A nested optimization method for an accelerating control plan of an aeroengine uses an inner loop and an outer loop to nest and optimize geometric mechanisms participating in the control of the aeroengine, wherein the geometric mechanisms participating in open loop control are globally optimized in the outer loop by constructing Bezier curves, the geometric mechanisms participating in closed loop control are locally roll optimized in the inner loop, the optimization of the inner loop is based on the outer loop, and the optimization of the outer loop takes the accelerating response time of the inner loop as an evaluation index. Preferably, the global optimization is performed on an outer loop by using a group intelligent optimization method, the individual dimension is determined according to the number of geometric mechanisms participating in open loop control