Gas Path Fault Tolerant Control of Aero-Engine Based on On-Board Model

WANG Cancan ,LI Zhengxin ,HUANG Jinquan

1 Beijing Machinery Research Institute,Beijing 100854 2 Key Laboratory of Aviation Power System of Jiangsu Province,Nanjing University of Aeronautics and Astronautics,Nanjing 210016

Abstract:A fault tolerant control method is proposed in this paper for a turbofan engine gas path degradation through the full flight envelope.A Quantum-behaved Particle Swarm Optimization (QPSO) algorithm is applied to obtain engine inputs adjustments,which contribute to construct off-line performance accommodation interpolation schedules.With a double closed-loop control system structure,command control is corrected based on real-time fault diagnostic results.Simulations indicate that fault tolerant control could reduce thrust and stall margin loss effectively in gas path faults.

Key words:aero-engine,gas path fault,fault tolerant control,QPSO algorithm

1 INTRODUCTION

Aero-engine gas path component faults account for more than 90% of total failures.There are many reasons for gas path faults,such as damage caused by foreign matter inhalation,mechanical fatigue fracture,corrosion scale and thermal creep.In order to prolong the service life of the engine,provide sufficient performance guarantees and complete the flight mission,degraded performance of the engine after failure must be restored.

At the beginning of the 21st century,NASA proposed to control a malfunctioning engine to ensure its safe landing.Since then,NASA formulated the Commercial Engine Damage Assessment and Reconfiguration (CEDAR) plan to ensure the thrust of malfunctioning engine by changing the control structure.Dan Ring proposed the concept of direct thrust control,which enables thrust feedback control for degraded engines.Chinese scholars have proposed an intelligent aero-engine performance degradation mitigation control technology,conducted digital simulation verification on a flight/push comprehensive simulation model,and obtained effective simulation results.However,none of the above approaches has looked at research on the method for recovery of surge margin,engine thrust has been limited by the low-pressure rotation speed in some simulation experiments and hence performance could only be partially recovered.

This paper selects suitable operating points and different operating conditions in the flight envelope,and calculates the adjustment value of the control parameters for gas path faults through the Quantum-behaved Particle Swarm Optimization(QPSO) algorithm,and constructs a performance recovery interpolation table.Finally,a dual closed-loop fault-tolerant control system is designed to ensure that the engine can still work stably and safely after a failure,and reduce performance losses such as thrust and the surge margin.

2 DESIGN OF ENGINE FAULT TOLERANT CONTROL SYSTEM

2.1 Questions

The efficiency degradation of a turbofan engine compressor is divided into three failure levels:low level (_1.50%),medium level (_3%),and high level (_5%).In order to demonstrate the failure performance recovery gas path component faults within the full envelope,a suitable flight operating point is selected as a reference point for design.Figure 1 shows the selection plan.A total of 16 design points and 8 test verification points were selected.Most of the design points were distributed on the boundary of the flight envelope and include various flight conditions such as take-off,climb,and cruise,having different performance requirements for the engine.

Figure 1 Design points and simulation points in fight envelope

Figure 2 Diagram of aero-engine gas path fault tolerant control system structure

Based on the 16 design points,the rotor speed and air flow were changed to improve thrust and restore the original surge margin by adjusting three control parameters:main combustion chamber fuel supply (

W

),exhaust nozzle area (

A

) and compressor guide vane angle (

CVGP

).

2.2 Acquisition of Control Adjustments for a High-level Fault

Under the condition that the compressor efficiency has a high-level failure (_5%) and the fan speed is 0.9 (based on the relative value of the design point) and 0.7,the QPSO algorithm was used to optimize the control parameters

W

,

A

,and

CVGP

to recover the engine’s performance indicators to the same state before the failure,the specific mathematical equation is as follows.

1) Performance index (fitness function):

2) Control parameter constraints:

3) Other constraints:

where

F

represents the engine thrust,

SMF

and

SMC

represent fan surge margin and compressor surge margin respectively;the subscript

nomal

is the physical quantity corresponding to no fault,the subscript

fault

is the physical quantity corresponding to failure,the subscript

opt

is the physical quantity corresponding to fault tolerant control;

n

,

n

,

T

represent low pressure,high pressure rotor speed and post-turbine temperature,

n

,

n

,

T

represent the corresponding maximum limit values respectively.

Since the value range of fitness function is between (0,1],the coefficients of the three performance indexes in Equation (1)are multiplied by 1/3,and different even power exponents are assigned to them to adapt to the different declined limits of performance parameters after the fault.In the QPSO algorithm,the number of particles is set to 20 and the maximum number of cycles is 50.

The operation realization process is as follows:with the research object of a turbofan engine component-level model,we calculate

F

,

SMF

,and

SMC

of the engine model under a certain flight state (height

H

,Mach number

Ma

,fan speed

n

) as well as

F

,

SMF,

SMC

with a high-level compressor fault;set the particle position vector in the QPSO algorithm as

X

=[

W

A

CVGP

] as the engine model input variable.Under the constraints of no over-temperature and no over-rotating,the optimization was performed according to Equation (1).Optimal position of the particle swarm was updated iteratively until satisfying QPSO algorithm termination condition,and 3 optimal control parameters are obtained.

The high-dimensional interpolation method was used to calculate the control adjustment of other working points in the flight envelope.Before that,linear regression analysis was conducted on the optimized control parameters based on the QPSO algorithm with fan speed divided into 0.9 and 0.7,and the mapping relationship between the control parameters and flight conditions was established as follows:

where Δ

W

,Δ

A

,Δ

CVGP

are control parameter adjustments required for performance recovery.

Based on the above mapping relationship,control parameter adjustments in the other flight states with engine high-level faults can be obtained by using the linear interpolation function in Matlab:

where

P

,

T

and

n

are flight conditions of design points;

P

,

T

and

n

are flight conditions of verification points;Δ

W

,Δ

A

,Δ

CVGP

are the adjustment values of each control parameters at the verification point.

2.3 Acquisition of Control Adjustments for Medium-level and Low-level Faults

For a medium level fault (_3%) and low level fault (_1.5%)of compressor efficiency,control parameter adjustments can be quickly calculated according to the following two equations:

where Δ

W

,Δ

W

,Δ

W

,Δ

A

,Δ

A

,Δ

A

,Δ

CVGP

,Δ

CVGP

,Δ

CVGP

correspond to the ad-justment of main combustion chamber fuel supply,exhaust nozzle area and compressor guide vane angle under the conditions of high,medium and low-level faults.

Through simulations,it can be verified that the adjustment amounts obtained with the above equations are approximately equal to that obtained with QPSO algorithm directly.For the compressor efficiency degradation between high,medium and low fault levels,high-dimensional linear interpolation is introduced to obtain the control adjustment.The interpolation table of performance recovery over the whole envelope was constructed,and control parameter adjustment values were calculated online according to the working state and fault level of the engine.

2.4 Design of a Closed Loop Fault Tolerant Control System

The structure of an aero-engine gas path component closed-loop fault-tolerant control system is shown in Figure 2,which is composed of internal and external double closedloop loops:the internal loop adopts LQG/LTR multivariable controllers to control the engine low-pressure speed and pressure ratio,so as to ensure that the engine can track the control command quickly,accurately and stably (the compressor guide vane angle command is controlled through the guide vane angle actuator).The performance recovery is implemented in the outer loop.Based on EKF filter,the fault diagnosis module monitors the engine health status in real time,provides fault information for the performance recovery interpolation table,and calculates the adjustment value of control parameters.Then,the control parameters are converted into the corresponding control command adjustment value through a nonlinear module,and the original command is corrected online to ensure the performance recovery of the engine after failure and reduce the loss of thrust and surge margin.

3 SIMULATIONS

3.1 Performance Recovery of a High-level Fault

The fan conversion speed is set at 0.9 at 16 design points,while the QPSO algorithm is used to solve the adjustment of control parameters.Figure 3 shows the numerical variations of thrust,fan margin and compressor margin,with the black line,red line,blue line correspond to normal condition,fault condition and performance recovery condition respectively.

The simulation results show that the engine thrust,fan margin and compressor margin were well restored by optimizing and adjusting the control parameters.Multivariable fault-tolerant control for the recovery condition with multi-performance index was realized.

3.2 Performance Recovery of Medium-level and Low-level Faults

Based on a specific design point,the compressor efficiency at [0.95,1] is divided into 0.005 intervals,the QPSO algorithm was used to solve the control parameters corresponding to different fault levels,and the change trend of adjustment can be seen in Figure 4.It can be seen that there is an obvious linear relationship between the adjustment value of control parameters and compressor efficiency,which demonstrates the effectiveness of Equations (7) and (8).

Figure 3 Performance accommodation of high-level fault case in design points with 0.9 fan speed

Figure 4 Control parameters adjustments with QPSO algorithm of different compressor efficiency in design point

Figure 5 Performance accommodation of medium-level fault case in design points with 0.8 fan speed

Figure 6 Performance accommodation of low-level fault case in design points with 0.8 fan speed

Figure 5 and Figure 6 show the performance recovery of the verification point in case of medium and low faults with a fan speed of 0.8 respectively.The simulation shows that the three performance indexes have been well recovered.The high-pressure speed and the temperature behind the turbine have increased after adjustment,but they have not exceeded the tolerance range.

3.3 Simulation of an Engine Closed-loop Fault-tolerant Control System

At the verification point

H

=2 km,

Ma

=1.2 with a high compressor efficiency fault (_5%),the closed-loop fault-tolerant control simulation was carried out with fan speed of 0.8.During the steady-state operation of the engine,the fault was injected at the third second.In order to compare the effects between fault-tolerant control and no fault-tolerant control,the external loop performance compensation loop was either disconnected or connected.

Dynamic performances of the simulation are shown in Figure 7,where the black line,red line,blue line correspond to normal condition,fault condition and performance recovery condition respectively.As can be seen from Figures 7 (a) -(c),compared with the no fault-tolerant curve,the three performance parameters under fault-tolerant control simulation have been recovered.The thrust value after fault tolerance was increased by 1%,the fan margin was restored to 98% of the original value,and the compressor margin was restored to 95%of the original value.Figure 7 (d) shows that turbine temperature increased by 5.4% with fault-tolerant control,but it was still within the limit value.Figures 7 (e) -(f) shows the simulation health parameters still contain noise after EKF filtering.After passing through a low-pass filter,each curve becomes stable and smooth,and high estimation accuracy is obtained,which is used to guide the outer loop to interpolate in real time and modify the engine control command.

Figure 7 Closed-loop fault tolerant control simulation results in high-grade case of compressor efficiency

4 CONCLUSIONS

This paper proposed a fault-tolerant control method to obtain performance recovery with an engine gas path fault and introduced the QPSO algorithm to calculate the adjustments of the design point control parameters off-line.Corresponding values of actual flight conditions are obtained based on high-dimensional linear interpolation.Through an engine double closed-loop fault-tolerant control system,control command parameters are modified online according to the fault diagnosis results.Simulation results show that this method could ensure the stable operation of the engine and reduce the performance loss of engine thrust and surge margin effectively.