Search

This article proposes a one-cycle digital twin method to simultaneously identify multiple component parameters in a power electronic converter (PEC), which only uses several sampling points (SPs) of voltage/current in one steady switching period. Consequently, only the open-loop digital model of PEC is required, and thus, high accuracy can be achieved since the additional linearized model of the controller or sampler is not needed. Besides, the digital replica is conveniently built with the aid of MATLAB/Simulink, which avoids complex mathematical modeling and, hence, is applicable to different converters. Based on the open-loop digital replica, particle swarm optimization (PSO) is employed to constantly update its component parameters, until its voltage/current data approximate that of the practical converter to complete the multiple parameters estimation. In the article, an example buck prototype is also built, which well validates that the proposed one-cycle digital twin can simultaneously identify parameters of inductor, capacitor, switch, and load, with high accuracy and convenience. Besides, it has good computational stability and is workable in different operating conditions and component degradations.

The past 20 years witnessed the invention of numerous converters by utilizing various topology derivation methods. Unfortunately, most of these methods are limited by pre-existing topologies or specific cells, causing the omission of some potentially valuable topologies. To break the limitations, a fundamental topology derivation method, namely components to converters (C2C), is proposed for nonresonant dc–dc converters. The basic idea of C2C is intuitively to derive topologies by combining separate components and filtering out valid combinations. Theoretically, C2C can derive converters more comprehensively since its results are not restricted by firm connections of the existing topologies or cells. However, C2C faces a heavy computing load caused by the massive combinations of components. Hence, a two-stage C2C topology derivation strategy is designed to alleviate the computing load. Furthermore, graph theory and dynamic programming are applied to computerize and optimize the above two-stage C2C. The two-stage C2C is utilized to derive single-switch two-port converters and single-inductor multiple-port converters. The derivation results show that all existing topologies with given components and numerous new topologies are derived automatically and simultaneously. Compared with the existing topology derivation methods, the proposed two-stage C2C is more thorough and automatic, facilitating more converters to meet various demands in practical applications.

The main challenges in designing the power reserve control (PRC) lie in the rapid estimation of the maximum available power ($P_\mathrm{avi}$) in real time and effective drift avoidance under the condition of fast-changing irradiation. Conventional PRC strategies utilize direct measurement or curve-fitting-based estimation to determine the $P_\mathrm{avi}$, which shows obvious limitations, such as extra hardware, implementation complexity, and slow estimation speed. Meanwhile, the drifted reserved power ($\Delta P$) may deteriorate the dynamic response performance, bring high dc-link overvoltage risk, and affect the system operation stability due to the voltage-step-$\Delta P$ regulation mechanism in conventional PRC methods. To address these issues, an improved PRC strategy with a fast $\Delta P$ transient mechanism is proposed with the rapid estimation of $P_\mathrm{avi}$ through just one pair of photovoltaic (PV) voltage and current sampling. Thus, the proposed algorithm is cost-effective, easy to implement, and compatible with existing PV systems since no additional hardware components are required. Moreover, the proposed PRC breaks through the inherent limitation of the voltage-step-$\Delta P$ mechanism, which can ensure the $\Delta P$ drift mitigation even under fast-changing weather conditions. Main experimental comparisons with other advanced PRC strategies were conducted to verify the advantages of the proposed algorithm in terms of rapid real-time available power estimation and drift avoidance.