Automated machine learning adjustment of the PID-controller for stability of the autonomous electronic systems under thermal cycling

To maintain the operations of autonomous electronic systems during rapid temperature changes, we employ machine learning methods to adjust the coefficients of the PID-controller. As we consider the problem of the influence of temperature conditions on the operation of optoelectronic devices, we propose a new method for solving it by usage of the PID-controlled Peltier element. We present the architecture of a model-based neural network we created to nonlinearly adjust coefficients of the PID-controller to manage the Peltier element. The effectiveness of the created tuning method is evaluated through a numerical experiment. Therefore, we prove an increase in the accuracy of the control of the working surface temperature by the use of the proposed method.