synthesis op• 資料種類:無 → table(僅由參數決定的運算子 —— 不接受影像或資料輸入)
• 呼叫:import acoustics; acoustics.synthesize_speed_ramp(rate=5000.0, duration=4.0, rpm_start=600.0, rpm_end=1800.0, orders=(1.0, 3.5), amplitudes=None, resonance_hz=None, noise_sigma=0.0, seed=None)(或 opsacoustics.get("synthesize_speed_ramp"))
升速過程:成分鎖定在軸階次上(可選一條固定在 Hz)。
> 以下的詳細說明為原文 —— 摘要與標題已翻譯。
Order tracking has no meaning at constant speed, so its ground truth needs a
signal whose shaft rate moves. Here the shaft rate ramps linearly from
`rpm_start to rpm_end` and each component's instantaneous phase is
`2 pi * order * revolutions(t)` — so it is locked to the shaft *exactly*,
by construction, and its order is known to machine precision.
`resonance_hz` adds one component at a fixed frequency instead. That is
the discriminating case: after angular resampling an order stays put and a
resonance smears, which is the whole diagnostic value of the transform.
Returns a dict, because a speed record without its speed profile is not
analysable: `signal (the waveform), rpm` (per-sample shaft rate),
`revolutions (cumulative, per-sample), rate, duration`,
`orders, total_revolutions, resonance_hz`, and
`max_component_hz`.
Raises `ValueError`: non-real / string / bool scalars, non-positive
`rate / duration / rpm_start / rpm_end`, an empty or non-finite
`orders, an amplitudes` of the wrong length, a length over
:data:MAX_SAMPLES, and **any component reaching Nyquist at the fastest
point of the ramp** — checked at `max(rpm)`, not at the mean, because a
ramp that is legal on average can alias at its top end and produce a
perfectly plausible spectrum.
• acoustic_condition_monitoring 族使用指南
• 範例資料目錄(下載 URL / 授權) —— 2-D 用 skimage.data(BSD/公有領域)加合成圖,3-D 給出真實資料源(Stanford/PDS 等)的下載 URL。
• 運算子來歷與參考文獻 —— 該運算子族所依據的研究/方法出處。
• 演算法的正典(作者・年份)與用途見上面的族使用指南。
• acoustic_condition_monitoring — py -3.11 examples/acoustic_condition_monitoring.py
table 作為輸入)synthesis)*Provenance: acoustics.py — ACOUSTICS 運算子登記表。本條目由 tools/opdocs.py md 自動產生(請勿手動編輯)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.