Resumen de Behavioral modelling and identification of power electronics converters and subsystems based on transient response
Virgilio Valdivia Guerrero
Nowadays, electrical engineers face significant changes in the way the electrical energy is generated and distributed to the consumers. On the one hand, the number of electronic and electrical loads in power distribution systems is continuously growing. Developments in power electronics technology during last decades have enabled the use of power-electronics-based subsystems as an alternative to mechanical, hydraulic and pneumatic subsystems, looking for more reliable and light systems, and a reduction in maintenance costs and environmental impact. On the other hand, due to the growth of alternative energy sources, power distribution systems supply the load not only from a single source but from a variety of energy sources such as batteries, fuel cells, solar panels and electromechanical generators. Consequently, power distribution systems are incorporating more and more power electronics converters, thus moving from traditional centralized architectures to distributed ones, where a variety of interconnected power converters supply a number of electrical and electronic loads with different voltage levels and dynamic requirements from a variety of energy sources. Current trends in power distribution systems for aircrafts, naval ships, hybrid/electric vehicles, telecommunications, datacenters, satellites as well as initiatives in micro-grids illustrate this concept. Such increase of power converters means increasing complexity of the power distribution architecture, at system-level rather than at converter-level. Dynamic interactions between regulated converters, activation of protections, connections and disconnections of load and sources are some problems to be faced by system engineers. Hence, modeling and simulation becomes a powerful system integration tool to ensure proper performance of the whole system at all operating conditions. However, modeling in power electronics have been traditionally focused on the design of the converters itself, rather than the integration of systems comprised of multiple converters. Most modeling approaches provide a detailed description of the internal signals of the power converter as well as requires detailed knowledge of its internal structure. However, new power distribution systems are comprised of a number of power converters provided by a variety of manufacturers. Companies need to protect their know-how, so they provide limited information about their products which is rarely sufficient to build a conventional average model or switching model. Also, excessively detailed models lead to unacceptable simulation time when large power distribution systems are analyzed. In order to cope with this lack of models, first proposals on system-level modeling of power converters have been recently proposed. The models are referred to as “behavioral models” since they only reproduces the behavior of the input-output voltages and currents and do not represent in detail the internal structure of the converter. Hence, they can be provided by the manufacturer while keeping confidential information. Moreover, behavioral models can be fully parameterized from a set of experimental measurements by the end user. However, the reported references so far are focused on DC-DC converters, either un-regulated or output voltage-regulated. The aim of this thesis is to propose novel system-level behavioral modeling and identification methods for several types of power electronics converters and other power-electronics-based subsystems typically integrated in power distribution architectures. The main characteristics of the proposed methods are the following ones: · The models are fully parameterized from a set of experimental tests and do not represent details about the internal structure of the modeled converter/subsystem. The models are simple, are built using dynamic transfer functions combined with nonlinear static functions, and reproduce the large-signal behavior of the converter/subsystem in terms of the signals required for system-level analysis, typically input-output voltage and currents. · The proposed identification method is based on the transient response of the input-output signals under a set of step tests. The tests are simple and can be carried out using low-cost equipment: switches, passive loads and a data acquisition system (e.g. an oscilloscope). From the transient response, a parametric identification algorithm is applied to identifiy transfer function models. --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
