Embedded Neural Network like PID Water Heating Controller Implementing Cycle by Cycle Power Control Scheme

Abstract

This paper experimentally investigates the heating process of a hot water supply using a neural network implementation of a self-tuning PID controller on a microcontroller system. The Particle Swarm Optimization (PSO) algorithm employed in system tuning proved very effective, as it is simple and fast optimization algorithm. The PSO method for the PID parameters is executed on the Matlab platform in order to put these parameters in the real-time digital PID controller, which was experimented with in a pilot study on a microcontroller platform. Instead of the traditional phase angle power control (PAPC) method, the Cycle by Cycle Power Control (CBCPC) method is implemented because it yields better power factor and eliminates harmonics in the power supply line. The smoothness of the heating process’s output response, which is a result of both empirical experiments and simulation results, demonstrates the efficacy of the suggested control mechanism, where the output response had a small ripple margin. The system performed according to design expectations and had unimpaired unity power factor throughout its operating range and no ripple was detected during its functioning.