Proper exploitation of information technology and the strengthening of the construction of waterway information are important ways to improve shipping efficiency and security. Considering the upstream reaches of the Yangtze River from Jiangjin to Chongqing (China) as a case study, a three-dimensional (3D) digital waterway system has been developed. The system, which comprises a server and a client, was designed to provide navigational and early warning services to shipping. In this paper, both the system architecture and the key methods of each module are presented. The functions and mutual relationships of the data input, database, 3D visualization, flow simulation, and early warning modules are described, and the key methods of the 3D digital waterway modeling, real-time simulation of waterway flow, and early warnings in navigation are discussed. The advantages of this system mainly concern the following aspects. (1) The 3D digital waterway model was established and integrated with virtual navigational elements, such as beacons, the water depth–contour surface, and navigable area, to enhance the intuitiveness of the digital waterway system and the comprehensiveness of its information. (2) Based on real-time hydrological data and a 2D hydrodynamic model, the waterway flow conditions with present-day inflow were simulated. The water depth distribution of a traditional inland electronic navigational chart (IENC) was provided, together with information on flow velocity and distribution, to improve the resolution and timeliness of the information acquisition. (3) According to Chinese inland navigational standards and the simulation of waterway flow conditions, the navigable area for shipping was zoned dynamically. In combination with the Global Positioning System (GPS) and networked ship-monitoring technologies, this system can provide early warnings for the safe navigation of shipping.

Centrifugal pumps are susceptible to various faults, particularly under challenging conditions such as high pressure. Swift and accurate fault diagnosis is crucial for enhancing the reliability and safety of mechanical equipment. However, monitoring data under fault conditions in centrifugal pumps are limited. This study employed an experimental approach to gather original monitoring data (vibration signal data) across various fault types. We introduce a multi–scale sensing Convolutional Neural Network (MS–1D–CNN) model for diagnosing faults in centrifugal pumps. The network structure is further optimized by examining the impact of various hyperparameters on its performance. Subsequently, the model’s efficacy in diagnosing centrifugal pump faults has been comprehensively validated using experimental data. The results demonstrate that, under both single and multiple operating conditions, the model not only reduces reliance on manual intervention but also improves the accuracy of fault diagnosis.

Underwater optical images often exhibit severe color distortion, weak texture, and uneven illumination due to light absorption and scattering in water. These issues result in unstable feature detection and inaccurate image registration. To address these challenges, this paper proposes an underwater image stitching method that integrates ORB (Oriented FAST and Rotated BRIEF) feature extraction with a fixed-ratio constraint matching strategy. First, lightweight color and contrast enhancement techniques are employed to restore color balance and improve local texture visibility. Then, ORB descriptors are extracted and matched via a KNN (K-Nearest Neighbors) nearest-neighbor search, and Lowe’s ratio test is applied to eliminate false matches caused by weak texture similarity. Finally, the geometric transformation between image frames is estimated by incorporating robust optimization, ensuring stable homography computation. Experimental results on real underwater datasets show that the proposed method significantly improves stitching continuity and structural consistency, achieving 40–120% improvements in SSIM (Structural Similarity Index) and PSNR (peak signal-to-noise ratio) over conventional Harris–ORB + KNN, SIFT (scale-invariant feature transform) + BF (brute force), SIFT + KNN, and AKAZE (accelerated KAZE) + BF methods while maintaining processing times within one second. These results indicate that the proposed method is well-suited for real-time underwater environment perception and panoramic mapping on low-cost, micro-sized underwater robotic platforms.

Underwater image stitching is a fundamental technology for marine environmental perception and underwater robotic vision, particularly for low-cost and micro-sized systems operating under resource constraints. However, light absorption and scattering, color distortion, low contrast, and limited computational resources significantly degrade the stability and embedded real-time feasibility performance of conventional stitching methods in underwater environments. This paper systematically reviews key technologies for underwater image stitching based on classical feature matching, covering image preprocessing, feature extraction and description, feature matching and geometric registration, and image fusion, with a focus on their applicability to low-power underwater robotic platforms. The principles, performance characteristics, and application scenarios of representative methods are analyzed. Based on this review, a dynamic Lowe-ratio–based feature-matching strategy is proposed and validated to address the loss of valid correspondences caused by fixed-ratio thresholds in low-texture and low-contrast underwater scenes. Experimental results demonstrate that, compared with fixed-threshold, cross-check matching, and SIFT-based methods, the proposed strategy achieves comparable or improved structural similarity and image quality while significantly reducing computational time on an embedded platform. The method is well-suited to low-cost underwater robotic platforms with limited computational resources and provides an efficient and robust solution for embedded real-time feasibility underwater image stitching.