The general rule is to use Lyaponuv candidature function, but in practice we can adjust the sliding mode gain using trial and error method as following :
- The bigs values increase the stability and cause high disturbance
- The small values reduce the high disturbance but waste the stability.
In principle, the gain for sliding mode control can be chosen according to the range of uncertain parameters as the main aim of using SMC is robustness. Especially, the gain K of K*sign(S) should be higher than maximum of the uncertainty to be suppressed while you show Lyapunov stability.
For order one, that is your case,you can use the proposition of Plestan et al [2011], however we make a generalisation for arbitrary order in Harmouche et al [2012] and Laghrouche et al [2014]. we propose an improvement in a submitted paper Chitour et al under revew.
Best regards.
Plestant et al [2011] https://www.researchgate.net/publication/232896347_New_methodologies_for_adaptive_sliding_mode_control
Harmouche et al [2012] https://www.researchgate.net/publication/261466137_Robust_and_adaptive_Higher_Order_Sliding_mode_controllers
Chitour et al (submitted) https://www.researchgate.net/publication/305186331_Practical_stabilization_of_perturbed_integrator_chains_with_unknown_bounds
Article New methodologies for adaptive sliding mode control
Conference Paper Robust and adaptive Higher Order Sliding mode controllers
Article Control of PEMFC Air-Feed System using Lyapunov-based Robust...
Article Practical stabilization of perturbed integrator chains with ...