Multi-GPU, Multi-Node Algorithms for Acceleration of Image Reconstruction in 3D Electrical Capacitance Tomography in...

Area of Science:

• Engineering
• Computer Science
• Physics

Background:

• Electrical Capacitance Tomography (ECT) is a valuable non-invasive technique for industrial process visualization.
• Traditional 3D ECT image reconstruction involves computationally intensive processes and large datasets, leading to significant time consumption.

Purpose of the Study:

• To propose and evaluate a novel parallel computing approach for accelerating 3D ECT image reconstruction.
• To enhance the efficiency and speed of data processing in 3D ECT systems.

Main Methods:

• Development of a distributed measurement system featuring a new parallel computing framework and an ECT-specific plugin.
• Implementation of multi-GPU, multi-node algorithms within a heterogeneous distributed system architecture.
• Performance evaluation using LBP and Landweber's reconstruction algorithms, focusing on data distribution and transmission.

Main Results:

• The proposed framework, integrated with a new network communication layer, substantially reduced data transfer times.
• Significant improvements in overall system efficiency were achieved through parallelized data processing.
• The parallel approach demonstrated a marked acceleration in the complex computations required for 3D ECT image reconstruction.

Conclusions:

• The developed parallel computing framework offers an effective solution for accelerating 3D ECT image reconstruction.
• This advancement enables faster and more efficient industrial process monitoring using 3D ECT.
• The multi-GPU, multi-node strategy is crucial for overcoming the computational challenges in high-resolution 3D ECT imaging.