Machine learning-based diagnosis methods have achieved remarkable success under the assumption that training and test data are identically distributed. However, a critical requirement of these is generalization capability to unseen domains when deployed actual scenarios. We introduce challenging problem domain generalization, i.e., learning from multiple source produce model can directly generalize without target information. adopt model-agnostic maximizes dot product gradients between...

Due to the material variations of lithium-ion cells and fluctuations in their manufacturing precision, differences exist electrochemical characteristics cells, which inevitably lead a reduction available capacity premature failure battery pack with multiple configured series, parallel, series–parallel. Screening that have similar overcome inconsistency among is challenging problem. This paper proposes an approach for lithium -ion cell screening using convolutional neural networks (CNNs)...

Domain generalization (DG) is a challenging task that aims to train robust model with only labeled source data and can generalize well on unseen target data. The domain gap between the may degrade performance. A plethora of methods resort obtaining domain-invariant features overcome difficulties. However, these require sophisticated network designs or training strategies, causing inefficiency complexity. In this paper, we first analyze reclassify into two categories, i.e., implicitly...

It is difficult to detect the surface defects of a lithium battery with an aluminum/steel shell. The reflectivity, lack 3D information on surface, and shortage many datasets make 2D detection method hard apply in this field. In paper, cross-domain few-shot learning (FSL) approach for lithium-ion defect classification using improved siamese network (BSR-SNet) proposed. To obtain critical battery, multiexposure-based structured light utilized. Then, heights cloud points are transferred...

Accurately predicting CO and CO2 content in blast furnace gas (BFG) holds immense significance, ensuring stable operation improving energy utilization. However, due to the variable operating conditions of (BF) ironmaking complex chemical reactions BF, it is difficult accurately predict changing trend BFG. To solve this problem, study proposes a temporal double graph convolutional network (TDGCN) model for prediction. It consists three parts: convolution, hypergraph TimesNet. Specifically, we...

Online dynamic prediction of the hot metal silicon content in blast furnace ironmaking process is crucial for stabilizing condition and improving molten iron quality. However, due to complex nonlinear correlations time-varying time lags between variables, a challenging task. To tackle problem, we propose novel method, called temporal hypergraph attention network (T-HyperGAT), which combined (HyperGAT) gated recurrent unit (GRU) network. Specifically, Hyper-GAT used capture high-order input...

In the realm of event prediction, temporal knowledge graph forecasting (TKGF) stands as a pivotal technique. Previous approaches face challenges not utilizing experience during testing and relying on single short-term history, which limits adaptation to evolving data. this paper, we introduce Online Neural-Symbolic Event Prediction (ONSEP) framework, innovates by integrating dynamic causal rule mining (DCRM) dual history augmented generation (DHAG). DCRM dynamically constructs rules from...

Multi-variate time series forecasting plays a crucial role in addressing key tasks across various domains, such as early warning, pre-planning, resource scheduling, and other critical tasks. Thus, accurate multi-variate is of significant importance guiding practical applications facilitating these essential Recently, Transformer-based models have demonstrated tremendous potential due to their outstanding performance long-term predictions. However, for often come with high complexity...

Traffic prediction is a crucial task in intelligent transportation systems, which can help achieve effective management and optimization of traffic congestion. However, due to the complexity uncertainty accurate has always been challenging problem. The specific challenge this how model dynamics along dimensions temporal spatial reasonable manner while respecting utilizing heterogeneity data. To address aforementioned challenges, paper proposes new Transformer-based approach for prediction....