Accurate pre-drilling mud weight window (MWW) prediction is crucial for drilling fluid design and wellbore stability in complex geological formations. Traditional physics-based approaches suffer from subjective parameter selection and inadequate handling of multi-mechanism over-pressured formations, while machine learning methods lack physical constraints and interpretability. This study develops a novel physics-guided deep learning framework integrating rock mechanics theory with deep neural networks for enhanced MWW prediction. The framework incorporates three key components: first, a physics-driven layer synthesizing intermediate variables from rock physics calculations to embed domain knowledge while preserving interpretability; second, a geological sequence-matching algorithm enabling precise stratigraphic correlation between offset and target wells, compensating for lateral geological heterogeneity; third, a long short-term memory network capturing sequential drilling characteristics and geological structure continuity. Case study results from 12 wells in northwestern China demonstrate significant improvements over traditional methods: collapse pressure prediction error reduced by 40.96%, pore pressure error decreased by 30.43%, and fracture pressure error diminished by 39.02%. The proposed method successfully captures meter-scale pressure variations undetectable by conventional approaches, providing critical technical support for wellbore design optimization, drilling fluid formulation, and operational safety enhancement in challenging geological environments.

The intrusive rock buried-hill reservoir is one of the main targets for oil and gas exploration in the offshore sedimentary basins of China. In order to discover the reservoir as early as possible for decision-making to save costs, reservoir interpretation with mud-logging data needs to be studied. Previous studies have showed that the homogeneity rocks reservoir can be well interpreted with mud-logging data u the K_b (mechanical specific energy ratio) method and the W_L (tangential power)-W_H (vertical power) intersection method. However, reservoir interpretation with mud-logging data for intrusive rock buried-hill has not been reported. The key steps or parameters of these two kinds of methods used for reservoir interpretation need to be modified for the vertical variation of intrusive rock buried-hill. Furthermore, confirming the interpretation of these two kinds of methods has not been reported. Taking the Y area in the Qiongdongnan Basin of the South China Sea as an example, the intrusive rock buried-hill can be divided into four zones based on its characteristics and genesis in descending order: the sand–gravel weathering zone, the weathering fracture zone, the inner fracture zone, and the base rock zone. The reservoir can be well interpreted when taking the M_SE (mechanical specific energy) geometrical mean of the base rock zone as a basic value to calculate K_b. The reservoir will also be well interpreted when W_H ranges from 0 to 4.5 MPa and W_L ranges from 0 to 99 MPa in the column while intersecting. The layers of the reservoir can be interpreted as K_b < 1 in the sand–gravel weathering zone and the weathering fracture zone. The K_b < 1 and effective intersection of W_L-W_H layers at the same time could be interpreted as a reservoir in the inner fracture zone and the base rock zone. After combining the K_b method with the W_L-W_H intersection method, the reservoir of intrusive rock buried-hill can be well interpreted. The total thickness of the uninterpretable reservoir ratio is less than 20% compared to reservoir interpretation with well-logging for each well.