Hi @Thang
The Simulink file contains many components, including blocks from Stateflow and Simscape. Could you provide the same mathematical representation of the entire system that you intend to describe in a journal article, technical report, or your thesis? The requirement must be 100% the same!
We typically do not attempt to design everything or control multiple components within a single file. Instead, we divide a large system into manageable subsystems. Once all subsystems are controllable and their performances are satisfactory, we can integrate them into a single Simulink file.
My suggestion:
If the nominal model of the subsystem is available, either derived from first principles or estimated using system identification tools, then one should attempt to design a set of linear or nonlinear input signals that stabilize the subsystem and ensure that the subsystem's output behaves like the reference signal, 
.
.Assume that the nominal model can be expressed in this simple form:
.If the model is in transfer function, then convert it to state-space.
Design the input signal
,such that all subsystem states in the state vector 
are measurable. <-- This is the initial assumption!
are measurable. <-- This is the initial assumption!The model's dynamics is then forcefully transformed to
which can be rearranged to
.If the error signal is properly defined, then
guarantees that subsystem's output tracks the reference signal.
