In the digital transformation of the petroleum industry, the massive multi-source heterogeneous tabular data scattered across systems poses challenges to efficient management and intelligent applications due to its unstructured nature and complex inter-table relationships. Existing table retrieval methods mainly rely on keyword matching or inter-table structure analysis to handle explicit associations, but they show limited adaptability to queries involving implicit associations, resulting in constrained retrieval accuracy and result coverage in complex real-world scenarios. To address these challenges, this paper proposes an implicit association reasoning table retrieval method, Relatab, designed for unstructured tabular data. By jointly modeling query-table relevance and latent inter-table associations, it automatically identifies tables that are indirectly related to the query but crucial to the answer. Specifically, the method first employs semantic matching techniques to compute the relevance between the query and candidate tables, then mines implicit inter-table relationships through semantic similarity of table content and contextual connections, and finally fuses the two scores to optimize retrieval ranking, generating a set of tables that cover key information. Experimental results demonstrate that on the Spider, Bird, and CementingTables datasets, the Top-2 recall rates of the proposed method are 79.21%, 61.26%, and 77.88%, respectively, outperforming the baseline model DTR by 1.74, 2.33, and 1.85 percentage points. This validates the method’s ability to improve retrieval coverage in complex query scenarios and its applicability in the petroleum domain.