GANTRY CRANE CONTROL SYSTEM DESIGN USING ADAPTIVE PID CONTROLLER BASED ON LQR, GENETIC ALGORITHM AND Q-LEARNING

• Main Author: Wahyu Wijaya, Zulfa
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/52604
• Institution: Institut Teknologi Bandung

Gantry crane is an instrument that is often found in various industrial fields. In general, gantry cranes are used in industrial fields that require the transfer of heavy goods, such as for transporting containers at ports, waste at nuclear facilities, materials in building construction, and being used in factory automation. In operation, gantry cranes require an adaptive and optimal control system so that the system response in the form of trolley positions and load swing angle remains stable and has high accuracy in system conditions with or without disturbances, as well as varying load masses. In this research, a controller design is conducted in a simulation environment with MATLAB and Simulink software. The controller is designed by using Proportional Integral Derivative (PID) which is optimally tuned with Linear Quadratic Regulator (LQR) and Genetic Algorithm (GA). The conventional PID controller design is still considered not optimal because the PID gain parameter is not adaptive enough to changes that occur in the system but uses a fixed gain parameter value for various conditions which can be dangerous due to system non-linearity of gantry crane, uncertainty and disturbance. For this reason, the implementation of the Q-learning algorithm in the control system is carried out so that the PID controller can adaptively determine its gain parameters according to the system error in the form of position and angle errors at any time so that optimal results can be obtained. It can be seen from the test results on normal load tests, normal load tests with disturbances and variable load tests that the implementation of Q-learning algorithm in determining the PID gain parameter (QPID) provides better results than the conventional PID controller. In the tests, it is shown that the use of a QPID controller can reduce the position response settling time, position response overshoot and energy use of the controller.