A series of Ni-Fe/SBA-15 catalysts was prepared and tested for the catalytic hydrogenation of levulinic acid to γ-valerolactone, adopting methanol as the only hydrogen donor, and investigating the synergism between Fe and Ni, both supported on SBA-15, towards this reaction. The characterization of the synthesized catalysts was carried out by XRD (X-ray powder diffraction), TEM (transmission electron microscopy), H₂-TPD (hydrogen temperature-programmed desorption), XPS (X-ray photoelectron spectroscopy), and in situ FT-IR (Fourier transform–infrared spectroscopy) techniques. H₂-TPD and XPS results have shown that electron transfer occurs from Fe to Ni, which is helpful both for the activation of the C=O bond and for the dissociative activation of H₂ molecules, also in agreement with the results of the in situ FT-IR spectroscopy. The effect of temperature and reaction time on γ-valerolactone production was also investigated, identifying the best reaction conditions at 200 °C and 180 min, allowing for the complete conversion of levulinic acid and the complete selectivity to γ-valerolactone. Moreover, methanol was identified as an efficient hydrogen donor, if used in combination with the Ni-Fe/SBA-15 catalyst. The obtained results are promising, especially if compared with those obtained with the traditional and more expensive molecular hydrogen and noble-based catalysts.

In recent years, churn rates in industries such as finance have increased, and the cost of acquiring new users is more than five times the cost of retaining existing users. To improve the intelligent prediction accuracy of customer churn rate, artificial intelligence is gradually used. In this paper, the bidirectional long short-term memory convolutional neural network (BiLSTM-CNN) model is integrated with recurrent neural networks (RNNs) and convolutional neural networks (CNNs) in parallel, which well solves the defective problem that RNNs and CNNs run separately, and it also solves the problem that the output results of a long short-term memory network (LSTM) layer in a densely-connected LSTM-CNN (DLCNN) model will ignore some local information when input to the convolutional layer. In order to explore whether the attention bidirectional long short-term memory convolutional neural network (AttnBLSTM-CNN) model can perform better than BiLSTM-CNN, this paper uses bank data to compare the two models. The experimental results show that the accuracy of the AttnBiLSTM-CNN model is improved by 0.2%, the churn rate is improved by 1.3%, the F1 value is improved by 0.0102, and the AUC value is improved by 0.0103 compared with the BLSTM model. Therefore, introducing the attention mechanism into the BiLSTM-CNN model can further improve the performance of the model. The improvement of the accuracy of the user churn prediction model can warn of the possibility of user churn in advance and take effective measures in advance to prevent user churn and improve the core competitiveness of financial institutions.

The utilization of CO₂, as a cheap and abundant carbon source to produce useful chemicals or fuels, has been regarded as one of the promising ways to reduce CO₂ emissions and minimize the green-house effect. Previous studies have demonstrated that CO₂ (or HCO₃^–) can be efficiently reduced to formic acid with metal Fe under hydrothermal conditions without additional hydrogen and any catalyst. However, the pathways and kinetics of the autocatalytic CO₂ reduction remain unknown. In the present work, the reaction kinetics were carefully investigated according to the proposed reaction pathways, and a phenomenological kinetic model was developed for the first time. The results showed that the hydrothermal conversion of HCO₃^– into formic acid with Fe can be expressed as the first-order reaction, and the activation energy of HCO₃^– is 28 kJ/mol under hydrothermal conditions.

Microalgae are promising feedstocks for third-generation bio-oil, yet direct hydrothermal liquefaction (HTL) produces bio-oil with high nitrogen/oxygen contents and low higher heating value. In this work, pretreated Spirulina was used as feedstock, with formic acid as an in situ H₂ donor. Two catalytic systems, including noble metal mixtures and non-noble Ni-Cu/ZrO₂ alloys, were constructed to investigate their regulation effects on bio-oil yield, elemental composition, and product distribution during algae HTL. When testing noble metal materials, a series of binary mixtures of Pd/C combined with Pt/C, Ru/C, and Rh/C outperformed monometallic catalysts. The optimal performance was achieved at a Pd/C to Rh/C mass ratio of 3:2, delivering a maximum bio-oil yield of 56.51 wt% and a minimum nitrogen content of 2.94 wt%. Compared with non-catalytic HTL, the bio-oil yield increased by 65 wt%, while the nitrogen content decreased by 49 wt%, surpassing that of monometallic Pd/C (43.67 wt%). To address the high cost of noble metals, a series of NiCu-ZrO₂ catalysts was synthesized. ICP and XRD verified the successful formation of Ni-Cu alloy active sites. Among all catalysts, Ni₁₆Cu₄-ZrO₂ possessed superior hydrodenitrogenation capacity over noble metal catalysts; the nitrogen and oxygen contents of the corresponding bio-oil declined by 32.75 wt% and 9.54 wt% relative to the blank group, respectively. GC-MS, FTIR, and TGA confirmed that bimetallic catalysis effectively degraded nitrogen-containing heterocyclic compounds, such as indole and aniline, and facilitated cracking of high-boiling heavy fractions into light bio-oil components. A comprehensive comparison demonstrated that binary mixtures of Pd-Rh noble catalysts are suitable for high bio-oil yield scenarios, while Ni₁₆Cu₄-ZrO₂ is low-cost and features outstanding denitrogenation efficiency.