The analysis of interwell connectivity plays a pivotal role in the optimization of the injection‐production relationship, the formulation of an oilfield development plan and the description of the remaining oil distribution. Nevertheless, conventional techniques for determining interwell connectivity, such as tracer analysis, interference well testing, and numerical simulation, are characterized by high costs and low efficiency. To address this challenge, an increasing number of artificial intelligence methods have been proposed. However, these methods often necessitate the availability of a substantial quantity of labeled data to facilitate model training, and sensitivity analysis is required to ascertain the interwell connectivity of the model training results. Furthermore, the majority of existing methods are unable to perform dynamic analysis of interwell connectivity. This paper proposes a dynamic calculation method of interwell connectivity based on an attention mechanism to address the aforementioned issues. The interwell connectivity is taken as the intermediate parameter of the neural network, with the interwell connectivity between each injection well and the current production well in the well group inverted by predicting the daily fluid production of each production well in the well group. The experimental results demonstrate that the interwell connectivity calculations presented in this paper are in accordance with the actual operational conditions observed in the oil field and the reservoir expertize of industry professionals. This indicates that our approch can effectively support dynamic decision‐making in the production process of oil reservoirs.

Water injection allocation is critical for maintaining pressure support in mature offshore waterflooding reservoirs, but its optimization is complicated by delayed injection–production responses, interwell interference, limited intervention windows, and incomplete field labels for injector–producer connectivity. This study proposes a connectivity-aware optimization framework that couples an attention-based connectivity identification network, a group-level long short-term memory (LSTM) production surrogate, and particle swarm optimization (PSO). The methodological novelty lies in using prescribed connectivity labels in a field-informed semi-synthetic benchmark to quantitatively test whether dynamic injection–production sequences and static well-pair attributes can be transformed into interpretable connectivity estimates for injection allocation decision support. The benchmark contains five injectors, ten producers, daily injection and production histories, static well-pair attributes, response lags, and normalized connectivity coefficients generated under practical injection rate, lag, water cut, and adjustment constraints. The attention model recovered the dominant injector–producer relationships with MAE = 0.0146, RMSE = 0.0240, R² = 0.9835, cosine similarity = 0.9962, and top-three overlap = 100%. The group-level LSTM achieved MAE = 4.524 m³/d, RMSE = 5.963 m³/d, MAPE = 1.255%, and R² = 0.964 on the chronological test set. Across 15 optimization cases, the PSO module generated feasible injection reallocations under single-well rate, total-injection balance, and +/−15% adjustment constraints. The results should be interpreted as controlled methodological validation rather than direct field deployment; further testing with anonymized field data is required.