High-Frequency Resonance Analysis and Stabilization Control Strategy of MMC Based on Eigenvalue Method

Abstract

For the high-frequency resonant accident of modular multilevel converter (MMC) in real engineering, in this paper, the 10th order state-space model of MMC is established based on the dynamic phasor method. The state-space model of MMC controller for power control is established according to the practical engineering. Then the time delay of input and output is considered in the controller, and the Pade fitting in state-space model applicable to the eigenvalue method is introduced to the complete state space model of MMC. Based on the proposed model, the MMC high-frequency resonance phenomenon is analyzed by the eigenvalue method, and the accuracy of the model is verified by simulation results. Besides, it is demonstrated that high time delay causes high-frequency resonance in the MMC system based on the participation factor and the system eigenvalue trajectory. Then, the coupling relationship between the input and output time delay inside the controller and the d-q axis time delay is investigated, and a new time delay compensation method to suppress high-frequency resonance is proposed based on Pade fitting, which can be used jointly with the traditional method with good effect.

High frequency oscillation analysis and suppression strategy of flexible direct current system

“…According to Table 1, while the proportional coefficient of the current inner loop increases from 4 to 7, the delay stability margin of the MMC-HVDC system solved by the Lyapunov direct method significantly decreases from 396.35 µs to 232.47 µs, decreasing by 41.35%, and the risk of HFO caused by link delay also increases, which is consistent with the conclusion of references [14,18]. However, while the integral coefficient of the current inner loop increases from 125 to 185, the delay stability margin of the MMC-HVDC solved by the Lyapunov direct method slightly decreases from 321.05 µs to 320.72 µs, with the variation range of only 0.10%.…”

“…However, the Pade approximation introduces additional differential state equations, increasing the complexity of the model. Moreover, if various delays in the system are taken into account, the order of the state equations will become very high, leading to a "disaster of dimensionality" [14].…”