Gust Alleviation Control Using Robust MPC

Abstract

Disturbance suppression is one of the very important objectives for controller design. Thus, many papers on this topic have been reported, e.g. (Xie & de Souza, 1992; Xie et al., 1992). This kind of problem can be described as an H∞ controller design problem using a fictitious performance block (Zhou et al., 1996). Therefore, disturbance suppression controllers can be easily designed by applying the standard H∞ controller design method (Fujita et al., 1993). Disturbance suppression is also important in aircraft motions (Military Specification: Flight Control Systems Design, Installation and Test of Piloted Aircraft, General Specification For, 1975), and the design problem of flight controllers which suppress aircraft motions driven by wind gust, i.e. Gust Alleviation (GA) flight controller design problem (in short GA problem), has been addressed (Botez et al., 2001; Hess, 1971; 1972). In those papers, only the state information related to aircraft motions (such as, pitch angle, airspeed, etc.) is exploited for the control of aircraft motions. However, if turbulence information is obtained a priori and can also be exploited for the control, it is inferred that GA performance will be improved. This idea has already been adopted by several researchers (Abdelmoula, 1999; Phillips, 1971; Rynaski, 1979a;b; Santo & Paim, 2008). Roughly speaking, GA problem is to design flight controllers which suppress the vertical acceleration driven by turbulence. In the 1970s, turbulence was measured at the nose of aircraft (Phillips, 1971); however, the lead time from the measurement of turbulence to its acting on aircraft motions becomes very short as aircraft speed increases. Thus, the turbulence data which were measured at the nose of aircraft could not be effectively used. On this issue, as electronic and optic technologies have advanced in the last two decades, nowadays, turbulence can be measured several seconds ahead using LIght Detection And Ranging (LIDAR) system (Ando et al., 2008; Inokuchi et al., 2009; Jenaro et al., 2007; Schmitt et al., 2007). This consequently means that GA control exploiting turbulence data which are measured a priori now becomes more practical than before. Thus, this paper addresses the design problem of such GA flight controllers. If disturbance data are supposed to be given a priori and the current state of plant is also available, then controllers using Model Predictive Control (MPC) scheme work well, as illustrated for active suspension control for automobiles (Mehra et al., 1997; Tomizuka, 1976). However, in those papers, it is supposed that the plant dynamics are exactly modeled; that is, robustness of controllers against the plant uncertainties (such as, modeling errors, neglected 16