Diagnosing cavitation in industrial pipeline systems is difficult because acoustic responses vary greatly with operating conditions, and similar time‑frequency patterns may correspond to different cavitation states. To tackle this, we introduce POC‑VLM (Physical Operating‑Condition‑Guided Vision‑Language Matching).\ \ POC‑VLM first builds operating‑condition‑aware semantic descriptions by explicitly inserting sample‑specific variables into candidate cavitation texts. A time‑frequency‑conditioned text adaptation module then dynamically aligns these physical semantics with the observed acoustic spectrogram. The subsequent global‑local vision‑language matching stage captures both holistic patterns and localized discriminative cues.\ \ In parallel, a K‑Nearest Neighbours‑based non‑parametric operating‑condition prior estimates the physical plausibility of each candidate state, and this prior is fused with the vision‑language semantic evidence at the decision level.\ \ Experiments on three real‑world cavitation datasets show that POC‑VLM consistently outperforms representative data‑driven, multimodal, and domain‑specific diagnostic methods in terms of accuracy, recall, and other metrics. Ablation studies confirm the complementary roles of vision‑language semantic matching and operating‑condition priors, and demonstrate robust performance under varying sampling rates, noise levels, and observation durations.\ \ Review