Pulse Density Modulation Pattern Optimization using Genetic Algorithms

Abstract

Pulse Density Modulation (PDM) can be used to drive resonant power converters and is an alternative to Pulse Width Modulation (PWM). Its main advantage is simplicity, which allows a power device to achieve zero-current (or voltage) switching while performing load power regulation. Reduced switching stress hinders a converter from polluting power lines with electromagnetic noise. This technique is suitable for designing power converters that show a good overall power factor and low Total Harmonic Distortion (THD). PDM can be used to drive resonant (series or parallel) power converters. These converters are frequently used in induction heating applications where they are required to operate at high frequencies and deliver a wide range of output powers. Conveniently, the power factor produced by PDM converters is near unity and THD is low at high-output powers. However, at low-output powers, THD increases and the power factor gets far away from unity. This paper presents a technique that makes it possible to obtain optimal PDM patterns. Simulations are used to show that intelligent PDM pattern generation using Genetic Algorithms allows for an improved power factor and a reduced THD at low-output powers. A comparison with other PDM pattern generation techniques shows that AG patterns demonstrate a much better performance.