Actuator placement optimization for adaptive trusses using a two-level multipoint approximation method

Parameters in adaptive trusses considering placements of actuators, control gains as well as structure cross-sectional sizes, have been simultaneously optimized to suppress vibrations and to maintain stringent shape specifications. Since the placements of actuators are treated as (0,1) variables but the control gains and structural member sizes are continuous, the studied problem then becomes mixed (0,1)-continuous optimization, which is difficult to be tackled with conventional optimization methods since they are often encountered with local optimal solutions or numerous computational costs. A two-level multipoint approximation strategy combined with genetic algorithm, which proved to be applicable in truss topology-size optimizations involving discrete-continuous design variables, has been developed to address the current actuator placement problem. To enhance the efficiency of this method, a branched multi-point approximation function was introduced to construct the first-level approximation problem. Besides, a new fitness function as well as adaptive crossover and mutation operators were adopted to prevent the optimization process from too early converging to local optimal solutions. Numerical results were presented to illustrate the efficacy of this strategy in dealing with actuator placement problems.