科研成果

How does the capacity removal policy affect China’s economy? To quantify the policy outcomes and costs, a four-sector model with vertical market structures is built. The calibrated model shows that, to achieve the policy goal, 10% of equipment operation in the high energy-consuming sectors must be shut down. This policy leads to an improved energy structure in which total energy consumption drops by 4.75% at the cost of a contraction in economic growth, where the total output declines by 12.31%. The numerical experiments find that the optimal policy is to limit the production scale in both the iron/steel industry and the fossil energy industry, closing 9% and 7% of the production, respectively, since doing so minimizes output loss and improves the energy structure. This paper quantifies the impact of the current capacity removal policy and provides policy alternatives to reach the same policy target with a lower output loss.

Abstract Soluble and miscible states are two important thermodynamic states in academic research and practical applications but their quantitative distinctions are still fuzzy. In this study, for the first time, the mathematical formulations of the quantitative criteria for distinguishing the thermodynamic soluble and miscible states are analytically developed by means of the Flory–Huggins solution theory and solubility parameter. The quantitative bottom and upper solubility limits for a total of 13 binary and ternary mixtures are calculated at different conditions. Moreover, the composition, temperature and pore radius are specifically studied to evaluate their effects on the soluble and miscible states. On the basis of the work from this study, the insoluble, soluble but immiscible, and miscible states are definitively quantified and clearly distinguished for the first time.

In this letter, we report a quasi-vertical GaN Schottky barrier diode (SBD) fabricated on a hetero-epitaxial layer on silicon with low dislocation density and high carrier mobility. The reduction of dislocation is realized by inserting a thin layer with high density of Ga vacancies to promote the dislocation bending. The dislocation density is $1.6\times 10^8$ cm?2 with a GaN drift layer thickness of $4.5 μ \textm$ . The fabricated prototype GaN SBD delivers a high on/off current ratio of $10^10$ , a high forward current density of 1.6 kA/cm2@3 V, a low specific on-resistance of 1.1 $\textmØmega \cdot \text cm^2$ , and a low ideality factor of 1.23.

Degradation of phenols with different substituent groups (including –OCH3, –CHO, –NHCOCH3, –NO2, and −Cl) at boron-doped diamond (BDD) anodes has been studied previously based on the removal efficiency and •OH detection. Innovatively, formations of CO2 gas and various inorganic ions were examined to probe the mineralization process combined with quantitative structure-activity relationship (QSAR) analysis. As results, all phenols were efficiently degraded within 8 h with high COD removal efficiency. Three primary intermediates (hydroquinone, 1,4-benzoquinone and catechol) were identified during electrochemical oxidation and degradation pathway was proposed. More importantly, CO2 transformation efficiency ranked as: no N or Cl contained phenols (p-CHO, p-OCH3 and Ph) > N-contained phenols (p-NHCOCH3 and p-NO2) > Cl-contained phenols (p-Cl and o,p-Cl). Carbon mass balance study suggested formation of inorganic carbon (H2CO3, CO32− and HCO3−) and CO2 after organic carbon elimination. Inorganic nitrogen species (NH4+, NO3− and NO2−) and chlorine species (Cl−, ClO3− and ClO4−) were also formed after N- and Cl-contained phenols mineralization, while no volatile nitrogen species were detected. The phenols with electron-withdrawing substituents were easier to be oxidized than those with electron-donating substituents. QSAR analysis indicated that the reaction rate constant (k1) for phenols degradation was highly related to Hammett constant (∑σo,m,p) and energy gap (ELUMO - EHOMO) of the compound (R2 = 0.908), which were key parameters on evaluating the effect of structural moieties on electronic character and the chemical stability upon radical attack for a specific compound. This study presents clear evidence on mineralization mechanisms of phenols degradation at BDD anodes.

This paper presents a novel CFD-driven machine learning framework to develop Reynolds-averaged Navier-Stokes (RANS) models. The CFD-driven training is an extension of the gene expression programming method Weatheritt and Sandberg (2016) [8], but crucially the fitness of candidate models is now evaluated by running RANS calculations in an integrated way, rather than using an algebraic function. Unlike other data-driven methods that fit the Reynolds stresses of trained models to high-fidelity data, the cost function for the CFD-driven training can be defined based on any flow feature from the CFD results. This extends the applicability of the method especially when the training data is limited. Furthermore, the resulting model, which is the one providing the most accurate CFD results at the end of the training, inherently shows good performance in RANS calculations. To demonstrate the potential of this new method, the CFD-driven machine learning approach is applied to model development for wake mixing in turbomachines. A new model is trained based on a high-pressure turbine case and then tested for three additional cases, all representative of modern turbine nozzles. Despite the geometric configurations and operating conditions being different among the cases, the predicted wake mixing profiles are significantly improved in all of these a posteriori tests. Moreover, the model equation is explicitly given and available for analysis, thus it could be deduced that the enhanced wake prediction is predominantly due to the extra diffusion introduced by the CFD-driven model.

Neural coding is one of the central questions in systems neuroscience for understanding how the brain processes stimulus from the environment, moreover, it is also a cornerstone for designing algorithms of brain–machine interface, where decoding incoming stimulus is highly demanded for better performance of physical devices. Traditionally researchers have focused on functional magnetic resonance imaging (fMRI) data as the neural signals of interest for decoding visual scenes. However, our visual perception operates in a fast time scale of millisecond in terms of an event termed neural spike. There are few studies of decoding by using spikes. Here we fulfill this aim by developing a novel decoding framework based on deep neural networks, named spike-image decoder (SID), for reconstructing natural visual scenes, including static images and dynamic videos, from experimentally recorded spikes of a population of retinal ganglion cells. The SID is an end-to-end decoder with one end as neural spikes and the other end as images, which can be trained directly such that visual scenes are reconstructed from spikes in a highly accurate fashion. Our SID also outperforms on the reconstruction of visual stimulus compared to existing fMRI decoding models. In addition, with the aid of a spike encoder, we show that SID can be generalized to arbitrary visual scenes by using the image datasets of MNIST, CIFAR10, and CIFAR100. Furthermore, with a pre-trained SID, one can decode any dynamic videos to achieve real-time encoding and decoding of visual scenes by spikes. Altogether, our results shed new light on neuromorphic computing for artificial visual systems, such as event-based visual cameras and visual neuroprostheses.