Efficiently generating high-fidelity initial excitations remains a challenge for multi-input multi-output (MIMO) time waveform replication (TWR) of non-stationary signals. To address this, this paper proposes an adaptive multistep estimation (AME) algorithm that produces high-fidelity initial excitations capable of direct use in TWR, eliminating the need for iterative experimental corrections. The algorithm maintains robustness against modeling errors and system resonances during estimation. It employs a frame-based recursive architecture that avoids computationally expensive high-dimensional matrix operations. This architecture enables automatic regularization within each computational frame for effective error control. An excitation space optimization strategy is used to distinguish between excitations and noise, ensuring that only mild regularization is needed and thus preventing over-smoothing. Additionally, a selective regularization strategy is proposed that leverages prior information to apply regularization solely to components not captured by the prior knowledge, thereby yielding estimates closer to the true values. The algorithm also provides autonomous quality monitoring through cross-frame semi-norm comparison, eliminating dependence on expert intervention or experimental validation. Numerical simulations and experimental results validate the effectiveness of the AME algorithm in achieving high-fidelity initial excitation estimation for MIMO TWR of non-stationary signals.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号