Seismic Data Compression Using 1-Bit ADM and 1D CNN Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The massive amount of seismic data generated by distributed acoustic sensing poses challenges for efficient data handling and transmission, particularly in battery-driven wireless geophones, necessitating new methods for compression that do not burden the geophones with excessive processing demands.
Innovation Solution
A system utilizing 1-bit adaptive delta modulation (ADM) for in-field seismic data compression combined with a 1D convolutional neural network (CNN) at the data center for signal recovery, allowing for real-time implementation and reducing computational load on geophones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If complex compression algorithms are implemented at geophones, then data compression ratio is improved, but device complexity and energy consumption increase
Solution Approach 1:
The system divides the compression task into two segments: simple quantization at the geophone level and complex reconstruction at the data center level. This segmentation allows geophones to perform only lightweight operations while achieving overall high compression ratios through the distributed architecture.
Solution Approach 2:
The patent introduces an intermediary reconstruction network that operates at the data center to bridge the gap between simple quantized data from geophones and high-quality reconstructed seismic signals. This intermediary handles the complex computational tasks remotely, freeing geophones from processing burdens.
2Measurement precision
If more seismic sensors are deployed, then measurement precision is improved, but data transmission volume increases
Solution Approach 1:
The system extracts only the essential features of seismic signals by performing 1-bit quantization at the source, removing redundant information before transmission. This extraction approach maintains measurement precision while dramatically reducing the volume of data that needs to be transmitted and stored.
Solution Approach 2:
The patent changes the parameter representation from high-precision floating-point values to 1-bit quantized values, fundamentally altering how seismic data is encoded. This parameter transformation enables dense sensor deployment by reducing per-sensor data volume while preserving essential signal characteristics through the reconstruction network.
3Device complexity
If 1-bit quantization is used at geophones, then device complexity is reduced, but signal-to-noise ratio deteriorates
Solution Approach 1:
The system performs preliminary 1-bit quantization at the geophone level to simplify device operations, then applies preliminary reconstruction processing at the data center using the trained network. This two-stage preliminary action approach recovers signal quality after quantization, maintaining simplicity at the source while improving SNR during reconstruction.
Solution Approach 2:
The reconstruction network uses feedback from multiple quantized measurements to iteratively recover the original seismic signal with high SNR. The feedback mechanism allows the system to compensate for quantization noise by leveraging correlations across multiple sensors and time samples, restoring signal quality despite aggressive 1-bit quantization.
Data Source
AI summary
A system, method, and non-transitory computer readable medium for in-field seismic data compression. The system includes geophones each having a 1-bit adaptive delta modulation for converting analog seismic signals to a modulated signal and transmitting the modulated signal, and a data collection center having a 1D convolutional neural network for receiving and compressing the transmitted modulated signal. The 1D CNN is configured to suppress quantization and random noise in the modulated signal.