I. INTRODUCTIONFor enhancing control systems, there are two important information sources: sensors, which provide measurements of variables, and human experts who give linguistic instructions and descriptions about the system. Fuzzy logic controller (FLC) was created to combine these two different types of information by handling information coming from human operators. The main advantage of the FLC is that it can be applied to plants that are difficult to model mathematically [1]. Self-tuning of a FLC aims to adapt the controller to different operating conditions [2]. For successful design of a FLC, proper selection of input and output scaling factors and/or the tuning of other controller parameters, such as the representation and construction of the rule base or the determination of the position and shape of the membership functions, are conclusive jobs [2]. Basically, there are two different tuning approaches to achieve optimal parameters for a FLC: on-line and off-line tuning [2]. Off-line tuning scaling factor using genetic algorithm optimization method can be found in [3]. On-line tuning membership function using gradient descent optimization method is discussed in In this work, we introduce an auto-tuning mechanism for the scaling factors of a PD-type FLC. The gradient descent method is employed to optimally determine them on-line. This algorithm along with other classical controllers is tested experimentally using an inverted pendulum mounted on a cart. The inverted pendulum is a highly nonlinear and open-loop unstable system [13], [14]. Inverted pendulum system is often used as a benchmark for verifying the performance and effectiveness of control algorithms because of the simplicity of its structure [15]. Results show the effectiveness of the proposed control system. The paper is organized as follows. Section II introduces the mathematical model of the inverted pendulum. Section III gives a brief description about classical controllers which are experimentally tested in this work. In Section IV, we derive the auto-tuning algorithm for the scaling factors of the FLC. Section V describes the experimental setup. Section VI discusses the experimental results and Section VII offers our concluding remarks.
II. INVERTED PENDULUM MODELThis Section provides description of the inverted pendulum used in this study. Fig. 1 shows the free body diagram. Using Newton's second law, it can be easily shown that the dynamic equations of motion are as follow: The symbols used in (1) and (2) and their numerical values are defined (see Table I).An Auto-Tuning Method for the Scaling Factors of Fuzzy Logic Controllers with Application to SISO Mechanical System Gamal Abdel Nasser, Abdel Badie Sharkawy, and M-Emad S. SolimanInternational Journal of Materials, Mechanics and Manufacturing, Vol. 3, No. 1, February 2015 The linearized model equations areWe used equations (3) and (4) for solving the Riccati equation in LQR controller to get the optimal state feedback gains.
III. CLASSICAL CONTROL ALGORITHMS
A. LQR (Liner Quadra...