涡轴发动机稳态性能退化缓解控制方法

Steady State Performance Degradation Mitigation Control of Turboshaft Engine

关键词:  涡轴发动机  性能退化缓解  性能退化估计  最优控制  神经网络

Key words:  Turboshaft engine  Performance degradation mitigation  Performance degradation estimation  Optimal control  Neural network

摘要:

为解决涡轴发动机性能退化后引起的涡轮后温度升高、耗油率增加等问题，提出一种基于压气机导叶调节的涡轴发动机稳态性能退化缓解最优化控制方法，设计了稳态性能退化缓解双回路控制系统。在动力涡轮转速控制回路采用基于交替方向乘子法的模型预测控制方法，在性能退化缓解控制（PDMC）回路，设计了基于堆叠神经网络的涡轮后温度指令模型，基于深度神经网络的退化量和涡轮后温度估计器，基于BFGS拟牛顿方法的最优导叶角度优化求解器，减小退化带来的涡轮后温度升高。仿真结果表明，性能退化缓解控制回路有效恢复了退化带来的涡轮后温度升高问题，可以将升高25K的动力涡轮后温度恢复到未退化水平，同时有效缓解了性能退化带来的耗油率升高和压气机喘振裕度降低问题，验证了采用导叶调节进行性能退化缓解控制方法的有效性。

Abstract:

In order to solve the problems of turbine exit temperature rising and specific fuel consumption increasing caused by the performance degradation of turboshaft engines， an optimal steady-state performance degradation mitigation control （PDMC） method based on the compressor inlet guide vane regulation is proposed for the turboshaft engine， and a two-loop control system is designed for steady-state PDMC. The model predictive control method based on alternating direction method of multipliers is used in the power turbine speed control loop. In the PDMC loop， a stacked neural network-based turbine exit temperature command model， deep neural network-based degradation and turbine exit temperature estimators， an optimal guide vane angle optimization solver based on the BFGS are designed to reduce the temperature rise caused by performance degradation. The simulation results show that the PDMC loop can effectively recover the temperature rise of the power turbine after degradation， and can restore the power turbine rear temperature increased by 25K to the non degraded level. At the same time， the PDMC effectively alleviates the problems of increased specific fuel consumption and reduced compressor surge margin caused by performance degradation， which verifies the effectiveness of the PDMC method using inlet guide vane regulation.