Search

Zero-shot learning (ZSL) can diagnose unseen faults without corresponding training data, which has aroused the researchers' interest. However, a prevailing challenge in most existing ZSL approaches is their limited effectiveness in distinguishing similar unseen faults. This paper proposed a category-tree-guided hierarchical knowledge transfer zero-shot fault diagnosis (CTZSD) method, which is a coarse-to-fine zero-shot fault diagnosis framework to alleviate this problem. To embody the similarities between fault categories, the concept of fault category tree is proposed, for which a data-attribute collaborative tree construction mechanism (DATC) is designed. Rather than relying solely on semantic knowledge, DATC involves data, which carries richer information, to complement the category similarity evaluation. A hierarchical knowledge transfer zero-shot fault diagnosis mechanism (HKT) is subsequently developed, utilizing the established category tree to gradually narrow down the options, thereby promoting the recognition of similar unseen faults. The mechanism treats the diagnostic outcomes and model parameters from coarse-grained tasks as knowledge and transfers them to fine-grained tasks for guidance, realizing a coarse-to-fine diagnosis. Aiming at providing discriminative information to further distinguish similar unseen faults, attention modules are integrated within HKT. These modules assess attribute weight, thereby directing the model's focus toward the discriminative attributes of similar unseen faults. Experiments on a real TPP industrial process demonstrate that the proposed CTZSD outperforms other traditional ZSL methods in distinguishing similar unseen faults, improving the average accuracy by at least 19.7%. • Category tree construction involving data to complement similarity measurement. • Category-tree-guided zero-shot diagnosis method to classify similar unseen faults. • The effectiveness of our method is verified in a real thermal power process. [ABSTRACT FROM AUTHOR]

With the fuzzy boundary between normal and abnormal video data, which cannot be well distinguished by most methods, anomaly detection in video requires better characterization of the data. First we give a convolution-enhanced self-attentive video auto-encoder based on the U-Net architecture, which can extract richer image features. Secondly we design a dual-scale feature clustering structure for this encoder, which simultaneously compresses the channel and spatial structure features of the image to represent the features to obtain good coding characteristics and expand the boundary between normal and abnormal data. We also verify that our approach is equivalent to a class of auto-encoders for memory-guided learning. Finally, in the reconstruction task, since video auto-encoders are capable of triggering temporal time leakage phenomena that can lead to network performance degradation, we propose an anomaly score computation paradigm for video auto-encoders that utilizes the average frame anomaly score of a video clip to compute the first frame anomaly score in that video clip. Extensive experiments on three benchmark datasets show that our method outperforms most existing methods on large datasets with complex patterns. The code will be published at the following link: Anomaly-detection-guided-by-clustering-learning • We propose a self-encoder that uses a convolutional enhancement of the attention mechanism. • We proposes a two-scale depth clustering module based on K-means algorithm. The module works by constrained encoder, which is helpful to distinguish abnormal data from normal data. • We give an anomaly detection paradigm applied to the video encoder reconstruction task. [ABSTRACT FROM AUTHOR]

• N fate was compared in different application methods (furrow and hole) and depths (0–20 and 20–40 cm). • Furrow application at 20–40 cm (F40) increased N use efficiency and decreased N loss. • The N competitive ability of I du bamboo (1–2 year old) was significantly higher than that of the other ages. Appropriate fertilization practice is crucial to increase nitrogen (N) recovery efficiency and decrease N loss. Our aim is to determine the appropriate fertilization placement and the target age in moso bamboo forests. A field experiment was conducted to determine the effects of application methods (furrow and hole) and depths (0–20 and 20–40 cm) on the fate of urea-15N in moso bamboo forests. The 15N tracer technique was used to investigate the fates of the applied 15N fertilizer. The largest biomass was obtained from the bamboo culm (42.25–54.09%), but the maximum absorption and N content occurred in the leaves (28.86–39.31%). The N recovery efficiency of urea-15N was significantly higher for deep applications (30.62–31.14%) than for shallow applications (26.68–27.49%), whereas no significant difference was found between the furrow and hole applications. The residual 15N was mainly concentrated in the fertilized soil layer, and the residual 15N recovered from the 0–60 cm soil layer was 16.29–19.47%. The highest N recovery efficiency (31.34%) and lowest N loss (49.91%) were observed for the furrow application at 20–40 cm. In addition, the N competitive ability and N recovery efficiency of the I du (moso bamboo age, 1–2 years old) bamboo were 0.069 mg kg−1 d−1 and 13.29%, respectively, which were significantly higher than those of the other ages. Therefore, furrow application at depth of 20–40 cm is recommended for practical applications, and the I du bamboo should be the target of fertilization in moso bamboo forests. [ABSTRACT FROM AUTHOR]

• Ultrathin 2D Al-MOF nanosheets were synthesized by a solvothermal method. • The presence of a mixed solvent and sulfate ion is critical to the formation of the specific 2D Al-MOF. • Ther ultrathin morphology is beneficial for the exposure of active sites and mass/electron transport. • The optimized Al-MOF exhibited high catalytic activity with almost no overpotential and high selectivity for 2e− ORR to H 2 O 2. Electrochemical transformation from oxygen reduction reaction (ORR) to hydrogen peroxide (H 2 O 2) is one of the promising green techniques to synthesis H 2 O 2. However, due to the competing two processes of 2e− and 4e−ORR with different products, designing and synthesizing highly selective and active 2e− ORR electrocatalysts for H 2 O 2 production is fundamentally and practically important. In this work, we used a simple solvothermal method and self-template strategy to create ultrathin 2D Al-MOF nanosheets for efficient electrocatalytic synthesis of H 2 O 2 by 2e− ORR. The presence of a mixed solvent and sulfate ion is critical to the formation of the specific structure. Naturally, their ultrathin morphology and abundant mesoporous formation leads to the sufficient exposure of electrochemically active sites and high mass/electron transfer capability. Therefore, the Al-MOF NSs show excellent electrocatalytic performance for 2e− ORR. The optimized Al-MOF NSs-1 exhibited the best ORR catalytic performance with almost no overpotential and high selectivity for 2e− ORR to H 2 O 2 (94 %–98 %) in a wide range of low potentials in alkaline electrolyte. Meanwhile, the high reaction rate of 2.56 molg cat −1h−1 for H 2 O 2 production has also been achieved at 0.6 V. Furthermore, the Al-MOF NSs-1 enable to catalyze H 2 O 2 production with excellent durability and selectivity after long-term stability test (10,000 cycles) and 11 h of potentiostatic electrolysis. The present study indicates the promising application of metal organic frameworks (MOFs) materials in H 2 O 2 electrochemical synthesis. [Display omitted] [ABSTRACT FROM AUTHOR]