Genetic algorithm applied to programmable current adder

Summary form only given. Traditionally, pulsed power systems have been designed and operated by the synchronized trigger technology of switch, which guaranteed the output waveform a fast front, high voltage or current. Recently, in order to produce the output current in demand, Genetic algorithm (GA) has been introduced for the pulse shaping by optimizing the operation time of each switch and the charge voltage of each capacity. To investigate technical problems that would happen in full scale drivers, the characterization and modeling of a scaled down programmable current adder was studied in this paper, including how to build the circuit simulation model and associate it with the GA. Based on the modeling method called Partial Element Equivalent Circuit (PEEC), a pulse power supply simulation model with 20 R-L-C parallel modules and 118 current branches was established. GA program was used to optimize the output current waveform. The influence of each operator and parameter on the genetic algorithm optimization process was also studied. For instance, the mutation function of GA was improved, and the influence of calculation step on optimize process was studied. The feasibility of the algorithm was tested by optimizing output currents on a six-module circuit and the same research has been done on the whole twenty-module circuit. The most evident results showed that the output current well matched the required current, which indicated that the algorithm was effective, adjustable, and realizable. For next research, experimental setup was necessary and further verified the accuracy of simulation results.