Title : 
A gradient based recursive identification of mechanical systems with backlash-like hysteresis
         
        
            Author : 
Dong, Ruili ; Tan, Yonghong
         
        
            Author_Institution : 
Coll. of Mech. & Electron. Eng., Shanghai Normal Univ., Shanghai, China
         
        
        
        
        
            Abstract : 
It is known that backlash-like hysteresis is one of the non-smooth nonlinearities usually existing in mechanical systems. Due to the characteristic of non-smoothness, it is quite difficult to apply the traditional identification method to the modeling of dynamic systems with backlash-like hysteresis in mechanical systems. In this paper, a gradient based recursive identification algorithm for dynamic systems with backlash-like hysteresis is presented. In this method, based on the concept of Clarke sub-gradient, a so-called bundle method is used to obtain the optimizing search direction at the non-smooth points. Then, the gradient based algorithm is developed for parameter estimation. Finally, an example of applying the proposed method for modeling a simulated mechanical transmission system is presented.
         
        
            Keywords : 
control nonlinearities; gradient methods; identification; machine control; nonlinear control systems; nonlinear dynamical systems; power transmission (mechanical); recursive estimation; search problems; servomotors; Clarke subgradient concept; backlash-like hysteresis; bundle method; dynamic system modeling; gradient-based recursive identification algorithm; mechanical system; nonsmooth control nonlinearity; nonsmooth point; parameter estimation; search direction optimization; servo motor; simulated mechanical transmission system; Control systems; Displacement control; Fasteners; Gears; Hysteresis; Mechanical systems; Nonlinear dynamical systems; Optimization methods; Servomotors; Teeth;
         
        
        
        
            Conference_Titel : 
Control Applications, (CCA) & Intelligent Control, (ISIC), 2009 IEEE
         
        
            Conference_Location : 
St. Petersburg
         
        
            Print_ISBN : 
978-1-4244-4601-8
         
        
            Electronic_ISBN : 
978-1-4244-4602-5
         
        
        
            DOI : 
10.1109/CCA.2009.5281178