Fire-resistant oil boom is a special equipment used for emergency response to marine oil spill accidents. Its core function is to intercept floating oil while resisting high-temperature flames, preventing the spread of fire and oil pollution, and ensuring marine environmental safety. To verify the fire resistance performance of fire-resistant oil booms in actual fire environments, it is necessary to conduct continuous fire resistance tests to evaluate their structural integrity, thermal insulation efficiency, and flame retardancy. Based on standard fire resistance test requirements, this study designed and built a land-based water tank simulation test environment. The test site layout, flame loading method, temperature monitoring point arrangement, and data collection method were elaborated in detail, and the deformation, ablation, and failure processes of the oil boom under high-temperature combustion were systematically recorded. The test results show that the fire-resistant oil boom can still maintain key performance indicators in line with specification requirements under long-term flame action, verifying its fire resistance reliability. This study realizes the fire resistance test of fire-resistant oil booms under land-based water tank conditions for the first time, solving the problems of high cost and uncontrollability of marine tests. Meanwhile, through multi-parameter data monitoring, it provides data support for material optimization and failure mechanism analysis of fire-resistant oil booms, which has important reference value for improving marine oil spill fire emergency equipment.

Show more

In response to the dual carbon goals, industrial parks, as the main force of energy consumption, accelerating the substitution of green electricity is the key to emission reduction. However, the two-way fluctuations in the supply and demand of green electricity make it more difficult to predict the gap. The uncertainty of supply and demand makes it hard for traditional experience-based dispatching to accurately match supply and demand, which may lead to the abandonment of green electricity due to excess or the reliance on thermal power for supplementation. This not only affects the emission reduction effect but also increases energy consumption costs. Therefore, there is an urgent need for high-precision green power supply and demand gap prediction methods to support dispatching decisions. This paper proposes the DE-Transformer-BiLSTM regression algorithm. Firstly, correlation analysis and violin plot analysis are carried out, and then it is compared with various machine learning algorithms such as tree models, neural networks, linear regression, SVR and decision trees. The results show that the algorithm performs better overall in terms of mean square error, root mean square error, mean absolute error, mean absolute percentage error and coefficient of determination, providing effective technical support for industrial parks to precisely schedule the supply and demand of green electricity, improve the utilization efficiency of green electricity, reduce energy costs and promote the implementation of the dual carbon goals.

Show more

With the progress in chemo-biological imaging technology, accurate molecular site segmentation and ligand detection with rare data has become an essential mission in drug screening, protein engineering, and early-stage disease detection. Traditional deep networks require extensive annotated data and fail to generalize well with multiple modalities jointly. In this work, we present the Few-shot Multimodal Chemo-Biological Imaging Framework (CMA-FSL), which combines chemical mass spectrometry and biological microscopy by cross-modal attention and adopts metric-based prototypical learning to rapidly learn with only a few data points. Experiment results on the ChemoBio-FS dataset show that our model outperforms state-of-the-art methods in terms of Dice score, IoU score, and Top-1 accuracy with more than 6% improvement in 5-way 5-shot evaluation settings, with results of 82.4%, 76.8%, and 86.9% respectively, validating the practicability and efficacy of multimodal learning in chemo-biological imaging with only a few data points, and opening an new thought and paradigm on small sample molecular diagnostic and intelligence drug discoveries.

Show more

In large-scale power dispatching cloud environments, massive traffic dispatching suffers from low efficiency and insufficient observability. This paper proposes a seed-end-cloud collaborative intelligent dispatching and observability optimization scheme. A three-tier pipeline architecture based on a lightweight agent, a stateless processing cluster, and distributed storage is designed. Source traffic shaping is achieved through real-time data aggregation at the agent end and Zstandard compression technology. A server-led dynamic rebalancing mechanism and CNI network plugin traffic shaping strategy are used to address instantaneous traffic surges and uneven load distribution. Storage layer performance is optimized through a self-developed high-availability connection pool and a dual-trigger batch write mechanism. In a simulated 10,000-node power dispatching scenario, the system throughput reaches 503.7 thousand records per second, with end-to-end latency controlled within 115.6 ms, providing feasible support for high-concurrency, high-reliability power services.

Show more