A Neural Network-Based Real-time Casing Collar Recognition System for Downhole Instruments

TL;DR

This work addresses the challenge of accurate downhole positioning under signal degradation by introducing in-situ, real-time collar recognition using lightweight Collar Recognition Nets (CRNs) deployed on an ARM Cortex-M7 processor. A TPU-friendly approach combines a Temporal Convolutional Network backbone with depthwise separable convolutions, training on field CCL data to produce continuous collar probability maps that are post-processed into depth estimates. Key findings show that CRNs can operate with as few as $1{,}985$ parameters and $8{,}208$ MACs, achieving a maximum $F1$ score of $0.992$ (CRN-1) and $0.972$ (CRN-3) while maintaining real-time performance of $343.2\,\mu s$ per $1$ ms sampling interval on a Cortex-M7. This demonstrates the practical viability of autonomous, in-situ neural signal processing for downhole tools and paves the way for robust, low-power edge computing in oil and gas operations.

Abstract

Accurate downhole positioning is critical in oil and gas operations but is often compromised by signal degradation in traditional surface-based Casing Collar Locator (CCL) monitoring. To address this, we present an in-situ, real-time collar recognition system using embedded neural network. We introduce lightweight "Collar Recognition Nets" (CRNs) optimized for resource-constrained ARM Cortex-M7 microprocessors. By leveraging temporal and depthwise separable convolutions, our most compact model reduces computational complexity to just 8,208 MACs while maintaining an F1 score of 0.972. Hardware validation confirms an average inference latency of 343.2 μs, demonstrating that robust, autonomous signal processing is feasible within the severe power and space limitations of downhole instrumentation.

Figures (4)

• Figure 1: (a) Schematic cross-section of a typical oil and gas well structure. Representative casing collar signatures derived from magnetic response are illustrated in dark blue near the corresponding casing collars, while typical interference signals are illustrated in dark green. Adapted from xiao2025realizationxiao2025dataaugmented; (b) Photograph of an perforating gun, exemplifying a typical downhole instrument assembly used in oil and gas wells; (c) Schematic diagram of the internal structure of a downhole instrument; the battery is omitted for clarity; (d) Block diagram of the collar recognition system within the control capsule; (e) Progress flow diagram for casing collar recognition utilizing neural network and casing tally; (f) Deployment workflow for neural network models, illustrating the transition from a PyTorch model to an executable program.
• Figure 2: Schematic diagram of the AFE module within the collar recognition system.
• Figure 3: Network architectures of the Collar Recognition Nets (CRNs) proposed in this work. For clarity, the batch normalization and dropout layers following each convolutional layer or fully connected layer are omitted.
• Figure 4: (a) Ideal probability map derived from manually collar annotations; (b--e) Comparisons of probability maps and recognition performance between MAN xiao2025dataaugmented and CRN-1 through CRN-3 models; probability maps deviate more noticeably from the ideal map as network capacity decreases; (f) An example of full-length recognition results using CRN-3, demonstrating that the majority of collar signatures are correctly recognized.