Design and optimization of intelligent exploration logging instrument based on sonic logging technology

Abstract Acoustic logging technology plays an important role in energy exploration. In order to break through the limitations of traditional amplification methods, this paper proposes an intelligent acoustic logging system with hardware-algorithm collaborative optimization, and establishes an integrated intelligent processing closed loop of “waveform segmentation-feature extraction-physical constraint inversion”. The tool adopts the multipole array structure of 8 transmitters and 12 receivers, combines the Field-Programmable Gate Array+Advanced Reduced Instruction Set Computer Machine architecture to realize real-time data processing, and innovatively integrates the mixed model of U-Net 3 + Gated Recurrent Unit-Bayesian. By embedding Wyllie’s formula and Gassmann’s equation into the system, high-precision-head-wave detection and heterogeneous reservoir inversion are achieved. In the field test of 20 shale gas wells in Sichuan Basin, the scheme reduces the detection error of the first wave to 0.1 µs (reduced by more than 50% compared to traditional deep learning method), R² improved to 0.92 (significantly better than the typical value of pure data-driven model of 0.70–0.80), the signal-to-noise ratio to 28.3dB, and the interpretation time of single well data to 12 min. The signal-to-noise ratio is nearly doubled, and the interpretation time of single well data is shortened to 12 min, which is about 90% faster. The value of high precision, strong adaptability and fast response of the collaborative framework under complex geological conditions is fully verified, which provides a reference for the intelligent transformation of acoustic logging.