Dewi Anggraeni, Setyawan Purnomo Sakti, Agus Naba, Sugeng Rianto
Polymerase chain reaction, or PCR, is a scientific technique for examining DNA. The PCR process works by amplifying, which involves heating the material to a specific setting point. A thermoelectric heating element that is controlled to create the desired condition is used to carry out the heating. PWM (pulse width modulation) is used as a thermoelectric controller to obtain a different temperature response according to the applied voltage. In the tool system that is made, PWM will be used as an input variable because PWM will control the temperature of the thermoelectric, so the temperature will be the output value. A representation of artificial intelligence that can be applied in PCR temperature control systems is an artificial neural network (ANN). Several techniques of ANN, including the Levenber-Marquart (trainlm), the Bayesian regularization (trainbr), and the Scaled Conjugate Gradient (trainscg) algorithms, were compared for performance in this study. The results showed that the Levenber-Marquart method and the Bayesian regularization method both produced the best results in the PCR temperature control system, with an MSE of 0.951268 and 0.952529, respectively. Each algorithm also produced a fairly good regression value of around 0.99, while the scaled conjugate gradient method produced a fairly large MSE of around 2941 and a poor regression value of around 0.64. © 2025 Author(s).
Department of Physics, Brawijaya University, Malang, Indonesia; Sensor Technology Laboratory, Brawijaya University, Malang, Indonesia
