基于数据驱动的变循环发动机多变量自适应控制算法研究

Multivariable Adaptive Control Algorithm of Variable Cycle Engine Based on Data Driven

关键词:  变循环发动机  数据驱动控制  自适应控制  多变量系统  梯度下降法

Key words:  Variable-cycle engine  Data-driven control  Adaptive control  Multivariable system  Gradient descent algorithm

摘要:

针对变循环发动机包线范围大、自适应控制需求高的问题，在增广线性二次型调节器（ALQR）基础上，提出一种基于数据驱动的自适应ALQR（DA-ALQR）控制算法。构建了多变量系统自适应控制准则函数，以梯度下降法对控制器参数进行调整，借助基于数据驱动的动态线性化建模方法，递推进行输出对输入梯度的近似计算，实现了对ALQR控制参数的自适应调整。仿真结果表明，所设计的DA-ALQR控制器参数随发动机状态变化得到了有效调整，相较于ALQR控制算法，闭环系统动态性能得到了大幅度提升，推力在单外涵模式表现出较快的响应速度，在双外涵模式表现出更小的超调，转速同样表现出了超调减小和响应速度加快特性，验证了所提出方法的有效性。

Abstract:

To meet the requirement of wide flying envelope and satisfy the demand for high adaptive ability of the variable cycle engine （VCE） control system， a data-driven based adaptive augmented linear quadratic regulator （DA-ALQR） control algorithm which is based on ALQR control structure is proposed. The adaptive control criterion function of the multi-variable system is constructed， and the controller parameters are adjusted by gradient descent method. The data driven dynamic linearization modelling method is used to approximate the gradients of the outputs to the inputs recursively. The adaptive parameter adjustment of ALQR controller is realized. The simulation results show that the parameters of the DA-ALQR controller have been effectively adjusted as the engine state changes. Compared with the ALQR control algorithm， the dynamic performance of the closed-loop system has been greatly improved. The response of thrust is faster in the single bypass mode and achieves lower overshoot in the double bypass mode. The response of rotor speed also shows reduction on overshoot and response time. Simulation results verify the effectiveness of the proposed DA-ALQR method.