A Multitemporal Point Cloud Registration Method for Evaluation of Power Equipment Geometric Shape

Abstract

The 4-D evaluation based on light detection and ranging (LiDAR) point cloud data of power equipment geometric shape can accurately describe the evolution process of equipment deformation on the space–time scales and meet the needs of refined geometric shape evaluation of equipment. It is a new and increasingly important technology for the condition assessment of power equipment. As one key technology for 4-D evaluation of power equipment geometric shape, the multitemporal point cloud registration method needs to satisfy the requirements of high precision, high universality, and intellectualization. In this article, first, the multitemporal point cloud registration strategy based on the local invariant feature (LIF) was established. Second, the LIF extraction algorithm for point cloud based on convolutional neural networks (CNNs) was proposed, and the multitemporal point cloud registration method for power equipment was structured. Finally, experiments were carried out to verify the feasibility and performance of the proposed registration method. The experimental results indicated that the proposed LIF extraction algorithm had excellent point cloud feature description ability, and the multitemporal point cloud registration method had great universality and robustness. The research could provide a reference for the development of geometric shape evaluation technology for power equipment.