Search

[Display omitted] • An innovative amine-based thermal energy storage (ATES) technology was developed. • ATES system utilize 84% of CO 2 reaction enthalpy for heat output. • The system obtained high-grade heat at ∼130 °C and energy density of 210 kJ/kg. • System energy efficiency of 70% was obtained via regeneration energy improvement. Thermal energy storage (TES) bridges the intermittence of renewable energy to generate steady and efficient clean energy. While amine-based CO 2 capture is a leading technology for CO 2 emission reduction, we innovatively employed the amine technology for long-term thermal energy storage, namely amine-based TES (ATES). The system utilises the thermally reversible amine-CO 2 reaction as the energy storage working pair, where low-grade thermal energy is used to break the amine-CO 2 chemical bond with the heat stored and released via exothermic amine-CO 2 reaction to produce high-quality thermal energy when required. Using benchmark absorbent monoethanolamine (MEA), we simulate and optimize the ATES system, and demonstrate that ATES exhibits desirable thermal energy storage performance with high-quality heat output up to 131 °C, high thermal energy conversion efficiency of >70%, moderate energy density of 210 kJ/kg and long-term storage, which outperforms the water-based sensible thermal storage technology. More importantly, the ATES is built on commercially viable amine process and heat exchange system, indicating its great potential for commercial applications, such as space heating and industrial waste heat storage. [ABSTRACT FROM AUTHOR]

[Display omitted] • Quantitative assessment of electron and hole transfer dynamics at anatase nanocrystalline TiO 2 / water interface. • TiO 2 conduction band electron, trapped electron and trapped hole absorption spectra were derived by comparing TA spectra. • Water reduction occurs with two single exponential components with difference in time scale of 6∼7 orders of magnitudes. • Water oxidation: Water oxidation reactions occur multi-exponentially in 25 ns to 200 ms. • Pt nanoparticle on the TiO 2 surface catalyses both water reduction and oxidation reaction. Quantitative assessments of electron and hole transfer dynamics with water on anatase nanocrystalline TiO 2 films were conducted by employing a series of transient absorption spectrometers. For water reduction reactions, both conduction band and trapped electrons decay with two different single exponential components with the difference in time scale of 6∼7 orders of magnitudes. The faster reaction occurs in 8–16 ns, while the slower component shows a lifetime of 0.1∼1.4 s. Pt nanoparticle deposition on the TiO 2 surface switches the slower single exponential reaction to a stretched exponential with an accelerated half lifetime of 2–7 ms, indicating that this slower reaction is limited by the electron trapping-detrapping movements inside the TiO 2. Water oxidation reactions occur multi-exponentially from 100 ns to 200 ms with a half lifetime of 10 μs. Pt on the TiO 2 surface catalyses water oxidation, occurring from 25 ns with a half lifetime of 8 μs. [ABSTRACT FROM AUTHOR]

Feature selection is a challenging and increasing important task in the machine learning and data mining community. According to different learning scenarios such as the availability of supervision information, feature selection can be broadly categorized as supervised, unsupervised and semi-supervised methods. Traditionally, feature selection algorithms are scenario-specific, i.e., different types of feature selection algorithms were proposed and investigated separately for specific learning scenario. In this paper, we aim to develop a unified view of supervised, unsupervised and semi-supervised feature selection methods. Based on the Hilbert-Schmidt independence criterion (HSIC), we first show that supervised, unsupervised and semi-supervised feature selection methods share the same objective. In discussing the connection between the unified framework and HSIC least absolute shrinkage and selection operator (HSIC Lasso), we find that the proposed framework not only has a clear statistical interpretation that minimum redundant features with maximum dependence on the target output variable can be found in terms of the HSIC, but also enables the global optimal solution to be computed efficiently by solving a Lasso optimization problem. Under this unified view, a new unsupervised feature selection algorithm is also proposed and demonstrated with several benchmark examples. • A unified view of supervised, unsupervised and semi-supervised feature selection methods is presented. • It has a clear statistical interpretation and enables the global optimal solution to be computed efficiently. • It is a useful abstract tool since feature selection can be investigated in a unified way. • A new unsupervised feature selection algorithm is proposed under this unified view. [ABSTRACT FROM AUTHOR]