From the controller design framework, a simple analytical model that captures the dominant behavior in the range of interest is the optimal. When modeling resonant circuits, complex mathematical models are obtained. These high-order models are not the most suitable for controller design. Although some assumptions can be made for simplifying these models, variable frequency operation or load uncertainty can make these premises no longer valid. In this study, a systematic modeling order reduction technique, slowly varying amplitude and phase (SVAP), is considered for obtaining simpler analytical models of resonant inverters. SVAP gives identical results as the classical model-order residualization technique from automatic control theory. A slight modification of SVAP, slowly varying amplitude derivative and phase (SVADP) is applied in this paper to obtain a better validity range. SVADP is validated for a half-bridge series resonant inverter and for a high-order plant, a dual-half bridge series resonant inverter giving analytical second-order transfer functions for both topologies. Simulation and experimental results are provided to show the validity range of the reduced-order models.
9 September 2013 in IEEE Transactions on Industrial Electronics
The wide variability of the load is the main problem of domestic induction-heating technology. The considered power control strategy should guarantee stability and regulation with a suitable response time and no overshoot independently of any perturbation or load variation. In addition to this, it has to ensure the operation under safety conditions. This paper is focused on the power control applied to one heating unit by means of continuous modulations. An inverse-based power controller is proposed for dealing with load uncertainties. The proposed power control combines different variable frequency pulsewidth modulation strategies (constant duty cycle, constant on-time, and constant off-time modulations) for reaching the target power level while minimizing the switching frequency. Finally, simulation and experimental results compared with other techniques in a prototype of an induction-heating system are presented in order to validate the proposed control strategy.
Resonant power converters are widely used in domestic induction heating (IH) cooktops due to their advantages in terms of size, efficiency, and power density. The wide range of different induction loads (with different electrical equivalent) and the wide range of required output powers, lead to the load identification as a mandatory and specially challenging aspect. This identification is needed to improve static operation, by identifying IH load that may lead to too high currents through the inverter, and dynamic operation, by correcting the gain of the digital controller according to the identified IH load. This work focuses on applying and evaluating the Particle Swarm Optimization (PSO) algorithm to identify the system load in domestic induction heating. Therefore, the obtained load parameters lead to a more efficient control strategy of the delivered power according to the user requirements. Simulation and experimental results in an induction heating prototype are performed in order to validate the proposed identification algorithm.
This paper analyzes the extended describing function technique (EDF) for modeling a resonant power converter applied to induction heating applications. Since in this specific application operating conditions may be far from resonance, this paper is focused on extending the model to several harmonics and verifying its validity range. The main goal is to obtain accurate control-to-power transfer functions which predict the resonant inverter small-signal behaviour. A simulation test bench is proposed and evaluated for measuring the open-loop duty cycle and frequency-to-power transfer function plots from time domain simulations of the switching model. Finally, the accuracy of the model is analyzed.