Recursive Least Squares-Based Adaptive Parameter Estimation Scheme for Signal Transformation and Grid Synchronization

Abstract

Utility-interfaced power electronic systems use a grid synchronizing framework, known as phase-locked loop and need a transformation of sinusoidal signals to rotating dq reference frame for control purposes. The voltage or current signal parameters, including instantaneous fundamental frequency, phase angle, and amplitude, need to be captured in the presence of harmonics, noise, and dc offset. This article proposes an adaptive estimation scheme for the same, using recursive least squares with time-varying covariance gains. Proposed adaptation of gain presents faster transient response and noise-tolerant steady-state response, achieving optimal tradeoff between the two. The covariance resetting mechanism is presented for a better dynamic profile during step changes in the signal. The scheme for the single-phase signal transformation is extended for transforming three-phase unbalanced sinusoids to the decoupled double synchronous reference frame. Stability analysis, design guidelines, and discrete-time realization of proposed methods are provided for reproducibility. Theoretical deductions of the proposed method are supported with several comparative test cases simulated in MATLAB/Simulink and the experimental results.