Research on the Reconfigurable Technique of SUAV Flight Control System

XU Xin-li

College ofMechanical Electrical Engineering andAutomation,

ShanghaiUniversity,

Shanghai, China

JIANG Zhen

College ofMechanical Electrical Engineering andAutomation,

ShanghaiUniversity,

Shanghai, China

FAN Bing-xiao

Police Fire Brigade,

Huangdao District of Qingdao City,

Qingdao, China

Abstract:The research on a Small Unmaned Air Vehicle(SUAV) has developed with various kinds of advanced technology, such as micro-machine technology, micro-motor technology, micro-electronics technology, integrated technology, and especially the nanotechnology.As the core of SUAV systems, the reconfigurable technique of flight control system can applie to the unknown fault and damage, and enhance the survivability and mission effectiveness of the aircraft obviously.It has become one of the attractive research subjects in modern aircraft integration design area.In this paper, several important methods for reconfigurable flight controlare introducedor analysed, andsome focus problems on the reconfigurable flight control technique are summarized and proposed. Finally, an outlook is did about the development tendency of reconfigurable flight control technique.

Key Words: SUAV; flight control system; fault and damage; reconfigurable technique

I.INTRODUCTION

The SUAV is also called flying robot. With small size, low cost, no casualties and the high survival ability of battlefield, therefore the research of SUAV has great significance in the military or civilian areas[1-3].During the aircraft flighting, its sensors, actuators, flight control computer and even the body itself will has fault or damage. And once the seriousness of the fault or damage to the flight control system of influence reaches a certain level, it will result in vehicle losing control and bring huge losses. So, it is hoped that the flight control system has the ability of "self-repairing", i.e., the ability of reconfiguration[4,5].On the basis of original control system and fault condition, the flight control system of reconfiguration can establish a new control system again, and guarantee system according to the original (or slightly reduce) performance index of safely operation to continue to complete the task or return safely[6-8]. The research is important to improve the safety and reliability, survivability and combat effectiveness of aircraft. And it has become one of the most popular research topics in the comprehensive design of modern aircrafts[9,10]. In this paper, based on a large number of literature references at home and abroad, the reconfiguration technology of the flight control system is discussed comprehensively.

II.PRESENTSITUATIONABOUTTHEDEVELOPMENTOFTHERECONFIGURATIONTECHNOLOGYOFTHEFLIGHTCONTROLSYSTEM

The reconfigurable control is developed on the basis of fault-tolerant control, involving fault diagnosis, control law reconfiguration, the system of timing matching, no traces of switching technology and so on. In 1982, the new concept of reconfigurable control was first proposed by the National Aeronautics and Space Agency(NASA)[11,12]. The research began in the self-repairing flight control (SRFC) system designed by the United States Air Force in 1984 and the reconfiguration plan of flight control system designed for commercial aircraft aviation[13].

The U.S. military has made some progress in some of methods, and conducted a number of flight test validation. From 1989 to 1990, the U.S. Air Force verifiedthe reconfiguration of control strategy based on pseudo-inverse method using F-15, and a comprehensive flight test and evaluation was conducted for all kinds of key technologies[14].It had achieved encouraging flight results, which marks the self-repairing flight control technology to start into the practical stage. In 1996, the U.S. Air Force conducted a flight validation for reconfigurable flight control systemusing VISTA/F-16[15], and it only did flight validation for a new reconfiguration strategy. In 1998,the U.S. Air Force used the unmanned X-36 tailless aircraft to take a reconfigurable flight test successfully usingadaptive control algorithm about neural networks for the first time[16]. The feasibility of reconfiguration of control law validation was conducted in the flight tests, while the robustness of flight reconfigurable system was evaluated in a fault state,which marked that the reconfigurable flight control technology had developed a new level.

During the “ninth five-year plan”, our country was also carried out research in the reconfiguration of flight control system, and some of ground semi-physical simulation tests were done for fault detection and isolation and control reconfiguration of reconfiguration system in case of a variety of rudder surface failure and damage. The phasic achievement had been gained, and a good foundation was laid for next step intelligent reconfiguration control technology. In early 2004, in China's first manned space project, the practical applicationwas firstly carried out successfully, such as rocket control system redundancy, spacecraft guidance and navigation system’s reconfiguration,whichshowed that substantial progresseswere made in the field of reconfiguration control technology in our country[17,18].

III.BRIEF INTRODUCTIONTO FLIGHT CONTROL SYSTEM RECONFIGURATION

Now, the flight control reconfigurable technology has attracted wide research because of its importance, and great progress has made. Researchers have made a lot of reconfigurable control methods. According to the situation of reconfigurable control which dependent on fault diagnosis or not, it can be divided into:

AIndirect reconfigurable control

If the reconfiguration control technology is designed according to fault detection and diagnosis information supplied by fault detection and isolation system andthe system dynamics model, it’s called the indirect reconfiguration control,such as pseudo-inverse method, multiple model adaptive control method and quantitative feedback reconfigurable control method. It mainly relates to fault diagnosis(including fault detection, fault isolation, fault identification) and reconfiguration of the control law. After a fault occurs, the fault diagnosis system diagnoses it and alarm, and the control law is reconfigured.

1)Pseudo-inverse method for reconfigurable control

Pseudo-inverse method is defined that when some control surfaces occur, it can use the remaining control surfaces without failure to combine according to the appropriate linear, and reorganize the aircraft's control signal. In mathematics, it usually multiplies a pseudo-inverse matrix in front of the original input matrix, so it’s called pseudo-inverse method. Pseudo-inverse reconfiguration method requires the accurate fault information of system. Using the information to modify feedback gain, it makes the control system performance after reconfigured maintain or close to the original performance.In 1980s, many scholars had done a lot of theoretical research and simulation for pseudo-inverse method used in the reconfiguration of flight control system, and conducted flight verification in F-15[5], indicating that is an effective reconfiguration control method.

Pseudo-inverse method is the first one of methods used in reconfigurable control, because it has simple structure and can be calculated and adjusted online for unknown fault. However, inversion and stability of the system is a key factor restricting its development.

2)Multiple model adaptive reconfiguration control method

Multiple model adaptive method is a primary method of adaptive reconfigurable control. It assume that a finite set can completely describe all sorts of possible damage of flight control system. Fews of identification modelsand corresponding controllers are set up in the flight control system, and it requires these controllers can guarantee robustness around the corresponding identification model in set of sufficiently large, which allows these sets to overlap each other to ensure the existence of reconfigurable control design solution. The whole system is built on a observer designed for appropriate choice of each damage model.The observer runs parallelly to find the closest to current working model of state, and switches to the appropriate controller.

However, the overall robustness of system is difficult to assess after use of the method, and the choice of switching order which ensures that the signal andsatisfactory performance of entire system have boundedness. It must be also considered that how to choose the design model and controller and when switch the model and controller in practical application.

3)Quantitative feedback reconfigurable control method

A quantitative feedback theory is used to design reconfigurable control laws in the method, and it is a method of off-line design and online choice. The design result is a set of fixed controllers. Since the controller can allow that object parameters has biguncertain,it ensures that the design result has strong robustness. The main feature of the method is to make filter and compensator been in series,which ensures design results to have robustness, and the system fault don’t need to identify.

The benefit of indirect reconfigurable control lies thatthe design of control law is relatively easier due to fault model. If the fault diagnosis is accurate, most of the reconfigurable control laws which is designed meet actual needs. However, this method also has disadvantages: Because the indirect reconfigurable control algorithm is serial in nature, it needs to make diagnosis, and then the control law reconfigurationwill be done. Soit needs to compromise on real-time and accuracy, which has brought considerable difficultiesfor an indirect reconfigurable control in practical application.

BDirect reconfigurable control

The reason why it’s called direct reconfigurable control is that the fault detection and diagnosis information is gained not through fault detection and isolation system, but the control law is designed dynamicly through online real-time parameter identification aircraft, such as: direct adaptive reconfigurable control method, feedback linearization control method andadaptive reconfigurable control method based on neural network.

1)Direct adaptive reconfigurable control method

The direct adaptive method is that a aircraft after failure meets conditions for positive realness through various construction techniques, following a direct adaptive control method in accordance with conditions of use. So it could use adaptive mechanisms to complete the reference model output followed by the fault plane.The U.S. Air Force had took a series of flight tests using direct adaptive reconfigurable control methods. The results show that it could complete reconfiguration control effectively and improve the system performance for gradual and abrupt failure, single or multiple faults, and a number of unforeseen circumstances.

The main feature of this approach is not necessary to know the failure information. As long as the system can be meet the conditions for positive realness, it can make the reconfiguration system stabilize, and achieve desired control effect. But its effectiveness is limited under certain conditions.

2)Feedback linearization reconfigurable control method

The basic idea is that the linear decoupling is done for nonlinear dynamics model of aircraft using feedback linearization, which is used to design reconfigurable control laws. Using feedback linearization method has the advantage to avoid designing and testing for a lot of gain scheduler, and reducestorage capacity requirements for the flight control computer.And it can adapt to more complex flight conditions.

3)Adaptive reconfigurable control method based on neural network

Adaptive reconfiguration method based on neural network is generally based onthe dynamic inversion control law,and a nonlinear flight control system is researched.After the failure/damage occurs, it adjusts the inversion error adaptively using online neural network. Thus, signal generated offsets error by fault, and it enhances instruction model to input for inverse controller, which makes a flight control system output accurately tracking the reference model output under fault conditions to achieve the desired dynamic performance, in order to ensure the aircraft still completes the mission safely after reconfiguration control. This needn’t to know the location or size of fault in advance, nor need for fault detection and identification, thus simplifying the design.

However, the learning speed of neural networks is slower generally. A fast learning algorithm must be studied to meet the need of real-time control. But how to ensure and evaluate the stability and convergence of control system after reconfigurable control is difficult problems of this approach.

The direct reconfiguration don’t dependent on the fault diagnosis, while the online control law reconfiguration is conducted directly. Thus,the real-time is better. But its disadvantage is difficult to design control law, whichis not an easy task to design a better adaptive control law for types of failure without relying failure.

IV.THE PROBLEMS EXISTING IN THE FLIGHT CONTROL SYSTEM RECONFIGURABLE TECHNIQUE AND ITS DEVELOPMENT TREND

AThe current problems

After years of development, reconfiguration control technology has been made a series of achievements in theory and application.But there is still a large number of theoretical and experimental work to be studied, and there are many difficulties, mainly in the following issues:

  • The question of reconfiguration control utility.The dynamics and structure fault of aircraft itself usually shows a serious nonlinear.Thus,the dynamics of fault plane is difficult to correctly describe for a small perturbation linear model, which makes the reconfiguration control method based on linear model fail. The coupling also exists in the system fault, and it is difficult to decouple, which also affected the application of reconfiguration Control.
  • The real-time problem of flight control law reconfiguration. As the SUAV’s flying environment is complex, it will be disturbed by multiple disturbances, such as wind, electromagnetic, and vibration.And system status and parameters varies very fast, which requires higher real-time about fault detection diagnosis and control law reconfiguration. Thus conventional algorithms can not meet requirements of real-time reconfiguration control, and the current intelligent reconfigurable methods based on neural networks and fuzzy inference also need to improve their real-time.
  • the question of robustness in reconfigurable control. In a reconfigurable control system, generally, because the control object is a nonlinear system, structure and parameters have strong coupling and time variability. The fault diagnosis method and reconfiguration controlmethod based on traditional model rely on the mathematical model of system excessively, which is very sensitive to modeling error, parameter perturbation, noise and interference.Thus,the robustness of diagnosis and reconfigurable control has become a serious problem.
  • the general problem of reconfiguration control. The existing reconfiguration control methods only study for a certain model aircraft or a control surface.They can not be applied to the reconfiguration control of other controlled objects, and the general application theories and methods about reconfiguration control have not yet formed.

BThe future trend of development

After almost two decades,the reconfigurable control system of flight control system has made ​​considerable progress, both in study of reconfiguration control system theory and flight validation of reconfiguration control methods. As for the current theoretical research and new technologies application at home and abroad, the focus development of flight control reconfiguration technology is to improve the adaptive capacity of flight control system furtherly for aircraft that is a complex system which has characteristics, such as nonlinear, strong coupling and time-varying.So the intelligenceof flight control reconfiguration technology will become the mainstream of future technology development.

In addition, with itself development of intelligent control theory, a variety of control methods from its cross-derived are gradually applied to the research about reconfigurable flight control system, such as the fuzzy neural network applied in the reconfiguration of flight control system. But a number of constraints about intelligent control theory itself should be seen, such as stability and other issues, which will be considered and resolved in design of reconfigurable control system.Additionally, a lot of simulation and flight test about results of existing studies still need to been conducted, which are used to study applications comprehensively under different flying mission. And so, the reconfiguration technology of flight control system will be pushed the practical application areas furtherly.

V.Conclusion

With development of science and technology, the domestic and foreign research for SUAV has made a number of encouraging progress, and people also have awareness of its inevitable development and wide application prospects. As the SUAV facing complex flying mission and flight environment, the function of its control system continues to be improved, and the role has become more and more prominent. The following requirements and attentions are increasing for the safety and reliabilityof flight control system. In order to improve the safety and reliability, combat survivability and effectiveness of aircraft, the reconfigurable flight control technology has become an important research direction aboutfuture design of flight control system, which has in-depth research value and broad application prospects.

REFERENCES:

[1]Juntong Qi, Jianda Han. Fault diagnosis and fault-tolerant control of rotorcraft flying robots: a survey[J]. CAAI Transactions on Intelligent Systems, 2007, 2(2): 31-39. (in Chinese).

[2]LilanZeng, Daobo Wang, Caigen Guo, Xianghua Huang. Survey of flight control technology for un-manned helicopter[J]. Control and Decision, 2006,21(4): 362-366.(in Chinese)

[3]Juntong Qi, XingangZhao, Zhe Jiang, Jianda Han. Design and implement of a rotorcraft UAV testbed[A].Proceedings of the 2006 IEEE International Conference on Robotics and Biomimetics[C].Kunming,China,2006.

[4]Heredia G, Ollero A, Mahtani R, BejarM, Remuss V, Musial M. Detection of sensorfaults in autonomous helicopters[A].Proc of the International Conference on Robotics and Automation[C].Barcelona,Spain, 2005.

[5]Hua D, Xing C, Xia Y. Fault diagnosis and faulttolerant control for wheeled mobile robots under unknown environments:a survey[A].Proceedings of theIEEE International Conference on Robotics and Automation[C].Barcelona Spain,2005.

[6]Xiaohui Qi,ZhijunYang,XiaobeiWu. Survey Study of Self-repairing Flight Control System on UAV[J]. Control Engineering of China, 2006, 13(6): 513-516, 549.(in Chinese)