Nonlinear Identification of A Wireless Control System: Comparison of NARX Model Results

Adnan ALDEMİR, Mustafa ALPBAZ Ankara University, Faculty of Engineering, Department of Chemical Engineering, 06100, Ankara, Turkey Abstract—This work has been carried out to comparison of two types of nonlinear models (wavelet network NARX and sigmoid network NARX) were applied for a process simulator that was used for the wireless control. Wireless input/output data obtained from the Cussons P3005 type process control simulator which three different temperatures (T2, T3 and T4) were selected as the controlled variables and the heater capacity was chosen as the manipulated variable. Wireless temperature experiments were achieved by using MATLAB/Simulink program and wireless data transfer during the experiments were carried out with radio waves at a frequency of 2.4 GHz. The two nonlinear models were developed with the aid of System Identification Toolbox of MATLAB using the data acquired from a unit step change on the heater capacity of the process simulator. The model orders used for the estimation of the model coefficients were determined with the aid MATLAB. According to the higher fit value and lower loss function observed in the case of the NARX model developed using wavelet network model has been found to be better than sigmoid network model. In addition, based on the simulation results, these two NARX models getting to the steady-state without any oscillations but wavelet network NARX models were discovered to be better than sigmoid network NARX models due to fastest rise time and fastest response time for this wireless control system. Keywords—System identification, MATLAB/Simulink, wavelet network NARX model, sigmoid network NARX model, wireless process control

[1]  Zhiqiang Zheng,et al.  Nonlinear system identification of a small-scale unmanned helicopter , 2014 .

[2]  Lennart Ljung,et al.  Nonlinear black-box modeling in system identification: a unified overview , 1995, Autom..

[3]  Francesco Corona,et al.  Regional models: A new approach for nonlinear system identification via clustering of the self-organizing map , 2015, Neurocomputing.

[4]  Keith Worden,et al.  Nonlinear system identification of automotive dampers: A time and frequency-domain analysis , 2009 .

[5]  Shan-Jen Cheng,et al.  Nonlinear modeling and identification of proton exchange membrane fuel cell (PEMFC) , 2015 .

[6]  Nithin V. George,et al.  Nonlinear system identification using a cuckoo search optimized adaptive Hammerstein model , 2015, Expert Syst. Appl..

[7]  Xiaojun Zhou,et al.  Nonlinear system identification and control using state transition algorithm , 2012, Appl. Math. Comput..

[8]  Süleyman Abdulwahab,et al.  Nonlinear Black-Box Modeling of a Reactive Distillation Process , 2012 .

[9]  Lennart Ljung,et al.  Regressor and structure selection in NARX models using a structured ANOVA approach , 2008, Autom..

[10]  Frank Allgöwer,et al.  Nonlinear structure identification of chemical processes , 1997 .

[11]  Mustafa Alpbaz,et al.  Black-Box Modeling of A Process Simulator Using Wireless Temperature Measurements , 2015 .

[12]  Erdal Kayacan,et al.  Nonlinear modeling and identification of an autonomous tractor-trailer system , 2014 .

[13]  Xiandong Ma,et al.  Nonlinear system identification for model-based condition monitoring of wind turbines , 2014 .

[14]  Lennart Ljung,et al.  System Identification: Theory for the User , 1987 .