关键词：控制器；并联有源；电力滤波器
摘 要：Shunt active power filters have been introduced as a way to overcome the power quality problems caused by nonlinear and reactive loads [9, 14, 23]. These power electronics devices are designed with the goal of obtaining a power factor close to 1 and achieving current harmonics and reactive power compensation [5, 6, 15]. The usual approaches [5, 15] for the control of shunt active filters are based on two hierarchical control loops: an inner one that assures the desired current and an outer one in charge of determining its required shape and the appropriate power balance as well. The control structure followed in this Chapter is the one in [7], in which the current controller is composed of a feedforward action that provides very fast transient response, and also of a feedback loop which includes an odd-harmonic repetitive control that yields closed-loop stability and a very good harmonic correction performance. In turn, the outer control law is based on the appropriated computation of the amplitude of the sinusoidal current network and, aiming at a robustness improvement, this is combined with a feedback control law including an analytically tuned PI controller. However, although the control system performance is very good, it shows a dramatic performance decay when the network frequency value is not accurately known or changes in time. For a better assessment of this issue this Chapter shows the experimental behaviors under constant and varying network frequency. This performance degradation is also presented and analyzed in terms of the THD, PF and cosφ. The Chapter organization is as follows. Section 7.1 introduces the plant, the control objectives and the two hierarchical control loops. Section 7.2 shows the odd harmonic controller designed for constant network frequency, also analysing the performance degradation through the THD, PF and cosφ. Section 7.3 details the results for the varying sampling time repetitive controller including the stability analysis by means of robust control theory. It is worth to say that the stability analysis using the LMI approach introduced in Section 3.3 can not be applied in this implementation since the size of the resulting matrices makes the problem computationally unsolvable. Section 7.4 and Section 7.5 present the implementation of the adaptive pre-compensation scheme and the robust design respectively and, finally, a second order HORC is applied in Section 7.6.