Modelling of Force Feedback Linearisation

In Chapter 3 the proposed feedback linearisation control was explained (see Figure 3.6), which was developed for the unified model of an MR damper. However, replacing the unified MR damper with basic or complex MR damper, the control algorithm has to be modified. Referring to Figure 5.6-(b), the input of MR damper is the control current (I),

5.2 Numerical modelling Numerical investigation

modified linearised control scheme is illustrated in Figure 5.6-(a).

Here, feedback control is being used to implement a semi-active force generator. The proposed control system uses measurement of the damper force to linearise the non-linear damping behaviour. Essentially, the controller gains B and G can be tuned so that the actual force closely matches the set point force. In Figure 5.6, this desired set-point force is chosen to be proportional to piston velocity, so that the MR damper behaves as a linear viscous device. fr desired d desired d

Figure 5.6: Models of force-feedback linearisation control; (a) Friction control model; (b) Current control model.

were presented in Chapter 4. These values are required to be modified for the friction control model, These are determined through extensive simulation testing on the MR damper, which led to the feedback controller gain B=0.6 N/N and the feedforward gain G=0.925 N/N. It is clear that, the both of the control model linearise the non-linearity of the damper with same feedback gain while the feedforward gain alters. This alteration implies the direct relationship between the desired control current and the desired friction force. This identity will be used to create the observer-based control model where the plant and observer uses the different MR damper model in the next section.

5.2.3 Observer Design For SDOF System

Referring to Equation 3.12, a linear observer model was designed (Figure 3.4) and this was extended to the non-linear observer model (Figure 3.5). By substituting the non-linear MR damper with the ’basic’, ’complex’ and ’unified’ models of MR damper in Figure 3.4, three types of non-linear observer is created. Having validated the unified model as a true representation of the system’s response (Chapter 4), afterwards this model will be used to design the plant of the system. The proposed three observer are shown in Figure 5.7 (basic MR model for observer), Figure 5.8 (complex MR model for observer), and Figure 5.9 (unified MR model for observer). The linearised feedback controller (current control model) is used to generate the desired control current (see 5.6-(b)), but which is converted to the friction force by the Current-Friction gain (I_FR) in case of basic observer or complex observer (Figures 5.7, and 5.8) is used. This gain is evaluated by using the direct relationship between the desired control current and the desired friction force as described previously. The I_FR gain is determined through extensive simulation testing of observer-based control system, which led to the Current-Friction, I_FR=1200 N/A.

5.2 Numerical modelling Numerical investigation Base excitation Error dynamics Actual output Estimated output Velocity Current Force Unified MR model k_p Spring stiffness k_p Spring Stiffness L* u Observer gain 1 s 1 s 1/m 1/m du/dt I_FR Current-Friction Velocity Control f riction Force Basic MR Model 1 Desired Control current

Figure 5.7: Simulated model of observer for SDOF mass isolation system (plant is de- signed with unified model of the MR damper while the observer is designed with the basic model of the MR damper).

Base excitation Error dynamics Actual output Estimated output Velocity Current Force Unified MR model k_p Spring stiffness k_p Spring Stiffness L* u Observer gain 1 s 1 s 1/m 1/m du/dt I_FR Current-Friction Velocity Control f riction Force Complex MR Model 1 Desired Control current

Figure 5.8: Simulated model of observer for SDOF mass isolation system (plant is de- signed with unified model of the MR damper while the observer is designed with the complex model of the MR damper).

Base excitation Error dynamics Actual output Estimated output Velocity Current Force Unified MR model Velocity Current Force Unified MR Model k_p Spring stiffness k_p Spring Stiffness L* u Observer gain 1 s 1 s 1/m 1/m du/dt 1 Desired control current

Figure 5.9: Simulated model of observer for SDOF mass isolation system (plant and ob- server are designed with unified model of the MR damper).

The key point of the observer design step is the decision of the observer gain matrix. In Chapter 3 it has been discussed that regarding to [154], for acceptable estimation of the states, the dynamic of the observer model has to be faster than the dynamics of the controller. The control force is generated by the MR damper, and because of the un- modelled friction inside the MR damper, the dynamics of the MR controller is already slower than the dynamics of a linear system. In other words, choosing the dynamics of the observer faster than the linear model will forward the poles of the observer far away from the poles of the controller (on the left side of the imaginary axis), which will cause an unwanted saturation of the estimated values [154]. In order to avoid this problem (referring to text book [154]), the observer gain matrix is chosen to be a little bit slower than the dynamics of the linear system so, it is going to be enough faster than the dynamics of the controller. By using the pole placement method the dynamics of the observer was chosen to be 25 percent slower than the actual passive system. In addition, different gains could be chosen based on system requirements, so this is not the best one.

5.2 Numerical modelling Numerical investigation L=    −0.0060 0.4375   