Guowei Li

Welding quality is directly related to the strength and service safety of ship structures. Due to complex imaging environments, severe noise interference, and diverse defect forms, traditional non-destructive testing methods for welds still have shortcomings in detection accuracy and automation. Considering that the formation of welding defects follows a distinct temporal sequence and exhibits specific physical characteristics, this paper introduces a multi-source information fusion strategy to jointly model image features and time-domain signals during welding. A welding defect detection model is proposed that combines the multi-source convolution enhancement module MS-SCM (Multi-source Split Convolution Module) with the information feature fusion module TA-WTM (Temporal-Aware Windowed Transformer). By extracting multi-source features in parallel to enhance the model’s ability to characterize defects of different sizes, and by introducing temporal information fusion and a sliding-window strategy within the attention mechanism, the model’s ability to model the defect space and temporal features is improved while reducing computational complexity. The study conducted experimental verification on multiple types of weld samples, including cracks, pores, incomplete penetration, and no obvious defects. The results showed that the proposed model exhibited certain advantages in detection accuracy and stability.

When ships conduct offshore operations in the ocean, they are subject to disturbances from natural factors such as sea breezes and waves. These disturbances lead to movements detrimental to the ship’s stability, especially heave movement in the vertical direction, which profoundly impacts the safety of shipboard facilities and staff. To counter this, the active wave compensation device is widely used on ships to maintain the stability of the working environment. However, the system’s efficiency and accuracy are compromised by the significant delay incurred while obtaining real-time motion signals and driving the actuator for motion compensation. To solve the time delay problem of shipborne wave compensation equipment in motion compensation under complex sea conditions, it is necessary to improve the ship heave motion prediction accuracy in an active wave compensation system. This paper presents a prediction method of ship heave motion based on the particle swarm optimization (PSO) and convolutional neural network–long short-term memory (CNN-LSTM) hybrid prediction model. The paper begins by establishing the ship heave motion model based on the P–M spectrum and slice theory, simulating the ship heave motion curve under different sea conditions on MATLAB. This simulation provides crucial data for the subsequent prediction model. The paper then delves into the realization method of ship heave motion based on PSO-CNN-LSTM, where the convolutional neural network (CNN) is used to extract the features of the input signal, thereby enhancing the multi-source feature fusion ability of the LSTM neural network model. The PSO algorithm is then employed to optimize the network structure and hyperparameters of the convolutional neural network. The experiments demonstrate that the proposed PSO-CNN-LSTM hybrid model effectively addresses the problem of predicting drift and boasts significantly higher prediction accuracy, making it suitable for predicting the short-term heave motion of ships. The data show that the optimized root mean square error (RMSE) value under level 5 sea conditions is 0.01265 compared to 0.01673 before optimization, and the optimized RMSE value under level 6 sea conditions is 0.01140 compared to 0.01479 before optimization, which demonstrates that the error between the predicted value and the actual value of the model decreases. This improved accuracy provides reassurance in the model’s predictive capabilities and lays the foundation for improving the accuracy of the motion compensation system in the future.

Stacker cranes play a significant role in modern logistics and warehousing systems, with their terminal pickup accuracy directly impacting the safety and reliability of equipment automation systems. This article introduces a method for predicting the accuracy of the stacker crane fork’s pickup position in the assembly process. Based on the concept of small displacement torsors, we conduct a structural analysis of the stacker crane to identify various error sources that affect the terminal accuracy of the stacker crane and establish corresponding small displacement torsors. Then, the assembly error transfer and accumulation of principal components are expressed through the Jacobian matrix, and the contribution of different tolerance values, such as the precision of the crane’s transmission components or the geometric tolerances of the separation surface, to cumulative errors, is determined. The method’s comprehensive approach provides clear and practical guidance for predicting the assembly accuracy of a stacker crane, instilling confidence in its applicability and effectiveness.