登入選單
返回Google圖書搜尋
Poursuite de Trajectoire Par Commande Non Linéaire Robuste Fondée Sur la Platitude Différentielle
註釋In this thesis, the concept of exact feedforward linearisation based on differential flatness is introduced. Instead of feedback linearising differentially flat systems, their structure is used to design nominal feedforward inputs which linearise the non-linear system under the knowledge of the initial condition. Moreover it is proven that a unique solution exists in the cases, in which the initial condition considered for the nominal feedforward is not equal but close to the right initial condition. The notion of exact feedforward linearisation based on differential flatness furthermore indicates the way, in which the aforementioned nominal feedforward should be combined with any kind of feedback to stabilise deviations from the desired trajectory. Stability of this control scheme can be demonstrated when using simple extended PID controls for the feedback part by considering a stability result from Kelemen. Thereby the two different tasks of designing the velocity of the desired trajectory and of choosing the controller coefficients have to be traded off carefully. This is important with respect to the practicability of the proposed methodology: in the case of a given flat non-linear system, for which there already exists a linear PID-like controller stabilising the system in the vicinity of an operation point, a non-linear nominal feedforward based on flatness plus the existing PID-like controller can achieve very good tracking or set point changes. In view of industrial applications the robustness analysis of the proposed methodology is introduced making use of results issued from interval analysis. A comparison of parametric robustness of exact feedforward linearisation based on differential flatness with exact feedback linearisation is presented. Thereafter the robustness analysis in the case of exogenous perturbations/time-varying parameters is established for exact feedforward linearisation.