Multi-Objective Design Optimization of Multicopter using Genetic Algorithm

Different research groups work together with different objectives in a single project. In Multicopter, Power Electronics has been assigned a task to maximize flight time whereas structural and inertial group requires light weight structure with minimum power consumption respectively. Selecting optimal readily available off the shelf components as per the mission requirement can be tricky job. It requires a trade-off between objective functions. There is significant amount of work done on multirotor using single objective optimization depending upon mission requirement but limited data is available for multiple objective optimization. One major drawback of integrating off-the-shelf components in optimization is that while optimizing one objective, the other objective may be blown out of proportions because of the same variable dependence. A multi-objective design optimization provides a pareto front which can really help the designer decide which variables to choose according to mission requirement. The pareto front actually demonstrates the trade-off between the objectives. The research aims to highlight the usability of genetic algorithm in multi-objective design optimization of multirotor with off-the-shelf components. Flight Time, power consumption and price are optimized simultaneously without payload. Mixed Integer Linear programming incorporates indexed or boolean variable. However the adaption of indexed variable into Genetic Algorithm is not completely straight forward and is therefore discussed in the paper. The variables are indexed as per the selection of parameters. These parameters are actuator (combination of propeller and motor), high efficiency DC battery and Airframe. The resultant score diversity, fitness function and Pareto fronts indicated fairly convergent and promising results. However, larger set of components offering more trade-offs between various fitness functions would definitely challenge this setup.