GENETIC ALGORITHM-BASED OPTIMIZATION OF PID CONTROL TO ENHANCE V/F SCALAR CONTROL PERFORMANCE OF INDUCTION MOTORS

• Main Author: Afrianto
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/83047
• Institution: Institut Teknologi Bandung

FTMD ITB and PT INKA are developing a roller-rig rail vehicle simulator for train technology development. The device requires a drive system that is not only reliable and precise, but also inexpensive. There are several choices of drive systems, including AC servo motors and AC induction motors. AC servo motors are expensive, and require specialized controllers that are also costly. Meanwhile, induction AC motors are a type of motor that is reliable and relatively affordable, but inaccurate. In order to achieve good accuracy, induction AC motors need to be operated using a closed-loop control strategy. In general, there are two types of control strategies to choose from, namely vector strategies and scalar strategies. Vector strategies offer high control accuracy, but require many sensors and complex programming. On the other hand, the scalar strategy is easy, cost effective and simple as it only requires one speed sensor. Unfortunately, the scalar strategy produces relatively low accuracy. Therefore, in this study, a scalar control strategy is developed using the GA method in order to obtain a reliable, inexpensive, and accurate induction AC motor speed control system. This research includes system modeling, model validation, computer-assisted control system optimization, and experimental performance testing. The model is developed with the aim of facilitating an easy and cost-effective optimization process through computer simulations. The model is validated using experimental data generated from previous research. Optimization of the control system utilizes Genetic Algorithms (GA) to obtain the optimal PID constants. These constants are then tested in simulation to predict the system's speed response. This prediction is subsequently verified with the speed response obtained from experiments. In this study, the overall system was successfully modeled. The model, when simulated on a computer, showed a speed response that matched the data from previous studies. This indicates that the model is valid. The model was then optimized through a computer simulation using the GA method. This optimization process resulted in the optimal PID constants. These constants were then tested in a closed-loop system, both numerically and experimentally. The closed-loop system using the GA method showed a significant improvement in speed accuracy. Compared to classical closed-loop methods such as ZN and CC, the GA method produced better control performance, characterized by lower overshoot and shorter settling times.