A Neural Network-Based Real-time Casing Collar Recognition System for Downhole Instruments
This work addresses the problem of accurate depth control for downhole operations like perforation in the oil and gas industry, representing an incremental improvement with domain-specific optimization.
The paper tackled the challenge of autonomous casing collar recognition in downhole environments by proposing Collar Recognition Nets (CRNs), achieving an F1-score of 0.972 with 1,985 parameters and 8,208 MACs, and demonstrating real-time performance with 1,000 inferences per second and 343.2 μs latency on an embedded system.
Casing collar locator (CCL) measurements are widely used as reliable depth markers for positioning downhole instruments in cased-hole operations, enabling accurate depth control for operations such as perforation. However, autonomous collar recognition in downhole environments remains challenging because CCL signals are often corrupted by toolstring- or casing-induced magnetic interference, while stringent size and power budgets limit the use of computationally intensive algorithms and specific operations require real-time, in-situ processing. To address these constraints, we propose Collar Recognition Nets (CRNs), a family of domain-specific lightweight 1-D convolutional neural networks for collar signature recognition from streaming CCL waveforms. With depthwise separable convolutions and input pooling, CRNs optimize efficiency without sacrificing accuracy. Our most compact model achieves an F1-score of 0.972 on field data with only 1,985~parameters and 8,208~MACs, and deployed on an ARM Cortex-M7 based embedded system using TensorFlow Lite for Microcontrollers (TFLM) library, the model demonstrates a throughput of 1,000 inference per second and 343.2 μs latency, confirming the feasibility of robust, autonomous, and real-time collar recognition under stringent downhole constraints.