Optimal input shaping for vibration control of a flexible manipulator using genetic algorithm

Abstract

This thesis presents optimization of input shaping technique for vibration control of a flexible robot manipulator using genetic algorithms (GA). This simulation work is designed in Matlab based environment with sampling frequency of 2 kHz and implemented on a Pentium 4 2.66 GHz processor and the responses are presented in time and frequency domains. The simulation can be repeated for different payload of the system varying from 0 to 100g. In this work, a single link flexible robot manipulator that moves in horizontal plane is considered. Modeling is done using FE method where the system is divided into 10 elements and the damping ratio of the system are deduced as 0.026, 0.038 and 0.040 for the first three vibration mode respectively. The input shaping technique is used to reduce vibrations in the system. This method requires estimated values of natural frequencies and damping ratios to generate impulse sequences. It is noted that the input shaping control technique is a better control technique compared to the bang-bang torque input control technique. It can be further optimized by using GA, by determining the optimal natural frequencies to cancel the resonance modes in the system and thus reducing the vibrations. For input shaping with genetic algorithm (ISGA) versus bang-bang (BB) and ISGA versus input shaping (IS), the percentages of vibration improvement in term of area representation is about 2720.03% and 28.57% respectively. In this work, GA optimization method not only reduces the vibrations, but also reduces time delay. GA can also be used offline or online to tune the system to achieve better performance due to modeling error. However, GA with input shaping technique increases the time and complexity of Matlab simulation.