Goegoes Dwi Nusantoro

Control of Pole-Climbing Robot Orientation using Self-Tuning Method

This paper describes the method to control a hybrid robot whose main task is to climb a pole to place an object on the top of the pole. The hybrid pole-climbing robot considered in this paper uses 2 Planetary PG36 DC-motors as actuators and an external rotary encoder sensor to provide a feedback on the change in robot orientation during the climbing movement. The orientation control of the pole-climbing robot using self-tuning method has been realized by identifying the transfer function of the actuator system under consideration, being followed with the calculation of control parameters using the self-tuning pole-placement method, and furthermore being implemented on the external rotary encoder sensor. Self-tuning pole-placement method has been explored to control the parameters q 0 , q 1 , q 2 , and p 1 of the controller. The experiments were done on a movement path in a form of a cylindrical pole. The first experiment was done based one the change in rotation angle of the rotary sensor with the angle values greater than 50˚ in the positive direction, whereas the second experiment was done with the angle values greater than -50˚ in the negative direction. The experiment results show that the control of the robot under consideration could maintain its original position at the time of angle change disturbance and that the robot could climb in a straight direction within the specified tolerance of orientation angle change.

System identification of a vacuum distiller using ANFIS with an ARMAX structure

As model complexity increased and unknown disturbance introduced, system identification became a useful method to solve necessity of the system model. This paper proposed an auto regressive moving average with exogenous input (ARMAX) structure system identification using Adaptive Neuro-fuzzy Inference System (ANFIS) for a Vacuum distiller. This vacuum distiller is used for bioethanol production. This approach differs from the conventional through the introduction of vacuum pressure disturbance as an exogenous input. Experimental results show that proposed method has comparable performance to the conventional Extended Least Square method.

System Identification using Modular Network SOM (mnSOM)

In this paper, we proposed a new identification method for changing parameter system based on modular network SOM (mnSOM). mnSOM is a generalization of Self Organizing Maps (SOM) formed by replacing each vector unit of SOM with function module. The modular function could be a multi layer perceptron, a recurrent neural network or even SOM itself. For system identification purpose, we employed ARX modules with recursive least square (RLS) learning. Inherit SOM capability of interpolating among function modules and well-known powerful capability of least square methods for system identification our RLS-mnSOM provides promising results for identification of changing parameter systems.