Machinery remaining useful life (RUL) prediction is an important task in condition-based maintenance. Data-driven methods have been widely studied and applied, however, almost all the researches learn degradation trends regardless of different fault conditions, which can lead to different degradation patterns. This article proposes a novel fault information assisted RUL prediction method based on a convolutional long short-term memory (LSTM) ensemble network, where fault conditions are obtained via fault knowledge transfer. Divergence minimization and domain adversarial adaptation are combined to transfer fault knowledge from a fault dataset to the run-to-failure data in a weakly supervised manner. With the predicted fault information, the RUL prediction network can learn various degradation patterns under different faults separately using a structure of multiple LSTMs. Then an ensemble strategy based on soft fault conditions is designed to get final RUL prediction results. Experiment on bearing datasets verifies the effectiveness of our proposed method.

Motor fault diagnosis is a fundamental aspect of ensuring the reliability of industrial equipment. However, industrial scenarios exhibit an inherent data scarcity problem, which imposes significant restrictions on the practical application of traditional deep learning-based intelligent fault diagnosis (IFD) methods. Typically, only a small volume of labeled data along with limited informative unlabeled data are available from industrial motors. Effectively utilizing informative unlabeled samples in the context of few-shot fault diagnosis poses a substantial challenge. In this article, a prototype refinement method for semi-supervised few-shot fault diagnosis based on deep reinforcement learning (DRL) is proposed. First, we propose to formalize a Markov decision process (MDP) of an iterative semi-supervised meta-learning strategy involving the selection of informative unlabeled samples and the refinement of category prototypes. Subsequently, we develop a mirror prototypical network (ProtoNet) structure for interaction with a DRL agent, which learns to adaptively select valuable samples to supervise the diagnosis process. Moreover, a state space involving feature embedding and category information is designed, and a comprehensive reward taking into account selection confidence, effectiveness, and representative is proposed. Extensive experiments on several motor experimental datasets verify the method’s effectiveness in few-shot diagnosis of unseen faults and new working conditions.

Machinery remaining useful life (RUL) prediction plays a pivotal role in modern industrial maintenance. Traditional methods entail the manual selection of useful features, which requires prior knowledge and lack adaptability to diverse cases. Moreover, as features may have different relevance to the degradation process at various stages, the prognostic performance will be limited by the utilization of fixed features throughout the full life-time. Additionally, most deep learning methods lack the perception of global information of features, which is critical to RUL prediction. To tackle these issues, an adaptive feature utilization method with separate gating mechanism and global temporal convolutional network (SGGTCN) is proposed in this paper. First, a separate gating mechanism is proposed to adaptively model temporal information within each feature individually through a series of designed separate gated residual modules. Second, an adaptive feature utilization method is proposed to evaluate and dynamically weight feature importance. Third, a global temporal convolutional network is proposed to model and fuse global temporal information for comprehensive sequential modeling. The effectiveness and superiority of the proposed method are validated by two prognostic case studies of turbofan engines and bearings.

The main bearing is the core part of the shield tunneling machine, and its health performance has a direct impact on the excavation efficiency and construction safety of the shield machine. To quantitatively evaluate the health performance of the main bearing, we propose a novel main bearing health assessment method based on two-stage signal decomposition. First, calculate the fault characteristic frequency (FCF) of each component based on the structure size and rotation speed of main bearing. Determine the FCFs with local maximum amplitude in the main bearing vibration signal's envelope spectrum. Then, the proposed two-stage signal decomposition (TSSD) method is applied to extract the optimized demodulation frequency band (ODFB) and each order fault frequency subsignal for each FCF. Finally, the energy ratio between the fault feature reconstructed signals and the original signal is used to construct the main bearing's health index. The first-level decomposition of TSSD is adopted to extract the ODFB filtered signal from the original main bearing vibration signal. The second-level decomposition is adopted to extract each order fault frequency subsignals from the envelope of the ODFB filtered signal extracted by the first-level decomposition. Finally, we use actual main bearing vibration signals of the shield machine to verify the proposed method and compare it with current health monitoring methods. The experimental results illustrate that the proposed approach can efficiently extract the fault features of the main bearing vibration signal and accurately evaluate the health performance of main bearing. The proposed method surpasses commonly used bearing health monitoring methods, such as Gini index and negative entropy. Moreover, the proposed TSSD can more effectively extract the fault features of the main bearing vibration signal than the current strong noise signal decomposition methods, such as two-level adaptive chirp mode decomposition (ACMD) and bandwidth-aware ACMD.

Accurate object detection is a fundamental component of autonomous apple-picking systems. In response to the insufficient recognition performance and poor generalization capacity of existing detection algorithms under unstructured orchard scenarios, we constructed a customized apple image dataset captured under varying illumination conditions and introduced an improved detection architecture, YOLO11-ARAF, derived from YOLO11. First, to enhance the model’s ability to capture apple-specific features, we replaced the original C3k2 module with the CARConv convolutional layer. Second, to reinforce feature learning in visually challenging orchard environments, the enhanced attention module AFGCAM was embedded into the model architecture. Third, we applied knowledge distillation to transfer the enhanced model to a compact YOLO11n framework, maintaining high detection efficiency while reducing computational cost, and optimizing it for deployment on devices with limited computational resources. To assess our method’s performance, we conducted comparative experiments on the constructed apple image dataset. The improved YOLO11-ARAF model attained 89.4% accuracy, 86% recall, 92.3% mAP@50, and 64.4% mAP@50:95 in our experiments, which are 0.3%, 1.1%, 0.72%, and 2% higher than YOLO11, respectively. Furthermore, the distilled model significantly reduces parameters and doubles the inference speed (FPS), enabling rapid and precise apple detection in challenging orchard settings with limited computational resources.