Search

This letter presents a variational Bayesian inference Neural Network (BNN) approach to quantify uncertainties in matrix function estimation for the state-space linear parameter-varying (LPV) model identification problem using only inputs/outputs data. The proposed method simultaneously estimates states and posteriors of matrix functions given data. In particular, states are estimated by reaching a consensus between an estimator based on past system trajectory and an estimator by recurrent equations of states; posteriors are approximated by minimizing the Kullback–Leibler (KL) divergence between the parameterized posterior distribution and the true posterior of the LPV model parameters. Furthermore, techniques such as transfer learning are explored in this letter to reduce computational cost and prevent convergence failure of Bayesian inference. The proposed data-driven method is validated using experimental data for identification of a control-oriented reactivity controlled compression ignition (RCCI) engine model.

This paper presents an integrated structure of artificial neural networks, named state integrated matrix estimation (SIME), for linear parameter-varying (LPV) model identification. The proposed method simultaneously estimates states and explores structural dependency of matrix functions of a representative LPV model only using inputs/outputs data. The case with unknown (unmeasurable) states is circumvented by SIME using two estimators of the same state: one estimator represented by an ANN and the other obtained by LPV model equations. Minimizing the difference between these two estimators, as part of the cost function, is used to guarantee their consistency. The results from a complex nonlinear system, namely a reactivity controlled compression ignition (RCCI) engine, show high accuracy of the state-space LPV models obtained using the proposed SIME while requiring minimal hyperparameters tuning.

Task transfer learning is a popular technique in image processing applications that uses pre-trained models to reduce the supervision cost of related tasks. An important question is to determine task transferability, i.e. given a common input domain, estimating to what extent representations learned from a source task can help in learning a target task. Typically, transferability is either measured experimentally or inferred through task relatedness, which is often defined without a clear operational meaning. In this paper, we present a novel metric, H-score, an easily-computable evaluation function that estimates the performance of transferred representations from one task to another in classification problems using statistical and information theoretic principles. Experiments on real image data show that our metric is not only consistent with the empirical transferability measurement, but also useful to practitioners in applications such as source model selection and task transfer curriculum learning.

The traditional sweeper dust reduction method is to spray water around the brush, spray electrostatic water and high-pressure water, because the ground is wet by water after the dust is easy to become mud, if the amount of water spray is not well controlled, less than the purpose of dust removal, more than the ground is not easy to sweep clean, the sweeping effect is not ideal, and there is a general problem of large water consumption. In winter, the temperature in many northern cities are below zero degrees, the water tank of clear water is easy to freeze, coupled with the traditional nozzle spray water volume, the suction nozzle can not be completely sprayed to the ground water suction pick up clean, easily lead to problems such as road ice, the formation of potential dangers to passers-by. For this reason, some domestic manufacturers, research institutes on this issue have been making unremitting efforts, and this paper describes the small road sweeping blowing dust truck as a newly developed economical and practical sweeping machinery.

This paper presents a framework to refine identified artificial neural networks (ANN) based state-space linear parameter-varying (LPV-SS) models with closed-loop data using online transfer learning. An LPV-SS model is assumed to be first identified offline using inputs/outputs data and a model predictive controller (MPC) designed based on this model. Using collected closed-loop batch data, the model is further refined using online transfer learning and thus the control performance is improved. Specifically, fine-tuning, a transfer learning technique, is employed to improve the model. Furthermore, the scenario where the offline identified model and the online controlled system are “similar but not identitical” is discussed. The proposed method is verified by testing on an experimentally validated high-fidelity reactivity controlled compression ignition (RCCI) engine model. The verification results show that the new online transfer learning technique combined with an adaptive MPC law improves the engine control performance to track requested engine loads and desired combustion phasing with minimum errors.

A copper-catalyzed S-methylation of sulfonyl hydrazides with TBHP was efficiently developed, providing a variety of methyl sulfones with good to excellent yields. The reaction can be carried out in water smoothly without any ligand or additive under mild conditions and this catalyst-in-water can be recycled several times.

A design for single-polarization single-mode photonic crystal fiber with rectangular lattice is proposed in this paper. The proposed fiber is studied by the full vector finite element method with perfectly matched layers. The single-polarization single-mode operation region of the fiber is achieved in a certain wavelength range with low confinement loss include the wavelength of 1.55 μm. The loss of one polarization is 0.124 dB/km at the wavelength of 1.55 μm and the confinement loss of the other one polarization is very high which can not ensure the transmission in the fiber. The single-polarization single-mode photonic crystal fiber is desirable for some polarization-sensitive applications such as high-power fiber lasers, fiber optic gyroscopes, current sensors and optical coherent communication systems.

A novel copper-catalyzed dual oxidative benzylic C H aminations of methylarenes with 2-aminobenzoketones in the presence of ammonium acetate was developed. This reaction represents a new avenue for 2-arylquinazolines with good yields. A key intermediate was detected and the kinetics isotope effect (KIE) indicated that C H bond cleavage was the rate-determining step.

We proposed a novel design of photonic crystal fiber with the characteristics of high birefringence and zero dispersion. The structure of the photonic crystal fiber is composed of the rectangular lattice with double-cladding and there is an elliptical air hole in the core of the photonic crystal fiber which has the same size with the holes of inner ring. Simulation results show that the elliptical air hole in the core can enhance the performance of mode birefringence and control the properties of chromatic dispersion at the same time. We obtain the high birefringence B = 2.1 × 10−3 and B = 2.9 × 10−3 at the wavelength of 1.31 and 1.55 μm respectively and the zero dispersion point at the wavelength of 1.31 μm. The proposed photonic crystal fiber with high birefringence and zero dispersion will be important in the applications of polarization maintaining transmission systems.

A broadband single-polarization photonic crystal fiber polarized filter based on surface plasmon resonance is proposed based on finite element method. Numerical simulations show the confinement loss of y-PCM (y-polarized core mode) is much higher than that of x-PCM (x-polarized core mode) in the wavelength range 1.20–1.63 μm. The confinement loss of y-polarized mode is 45,240 and 10,200 dB/m at the communication wavelength 1.31 and 1.55 um, respectively, and the corresponding loss of x-polarized mode is just 90 and 80 dB/m. When the fiber length is 3 mm, the bandwidth of extinction ratio better than -20 dB is greater than 430 nm covering almost all the communication wavelength. To our best knowledge, the bandwidth is the widest. The impacts of structural parameters on the resonance characteristics are also discussed. The structure could be further optimized for better result.

In this paper, a novel D-shaped photonic crystal fiber (D-PCF) based on surface plasmon resonance (SPR) is designed, and gold nanowire is used as the SPR active metal. A full-vector finite element method is applied to analyze the polarization filter characteristics. Phase matching between the second-order surface plasmon polariton (SPP) mode and the core-guided modes of x- and y-polarization can be met in two different wavelengths. By adjusting the diameters of the large air holes and the gold nanowire, this designed D-PCF can be used to the polarization filter in two communication bands. The loss of unwanted y-polarization mode can reach 244.9 and 292.8 dB/cm at the wavelength of 1.31 and 1.55 μm, respectively. With a length of 1 mm of the designed D-PCF, the bandwidths with the crosstalk better than 30 dB can reach to 88 and 150 nm, and the crosstalks are 208.4 and 249.5 dB the wavelength of 1.31 and 1.55 μm, respectively.

We propose and investigate a D-shaped photonic fiber refractive index sensor with rectangular lattice based on surface plasmon resonance. In such sensor, the nanoscale gold metal film is deposited on the flat surface where it is side polished. Numerical results show that the average sensitivity of Au-metalized surface plasmon resonance (SPR) sensor could reach as high as 8,129 nm/refractive index unit (RIU) in the dynamic index range from 1.35 to 1.41 as well as 2,000 nm/RIU from 1.33 to 1.35. Compared to conventional Au-metalized SPR sensors, the performance of our device is obviously better, and the production process is greatly simplified.

A kind of elliptical structure photonic crystal fiber with diameter increasing air holes is proposed. In this structure all the circular air holes are arranged in six elliptical rings. And the diameter of each rings increase from the inner to the outer. Moreover, we have choose some air holes to coating the metal Au and filling the liquid of the purified water. Properties, just like the confinement loss, of these structures with different diameters of the holes and thickness of the gold layer are numerical analyzed by using the finite element method. Numerical results show that the resonance strength in x-direction is much stronger than in y-polarized direction, and filling the liquid and changing the thickness of the gold film can significantly increase the loss of core mode polarized in the x-polarized direction.

Anthranilic acid amides are cyclized to the corresponding dihydro-quinazolinones in good yields with dicumyl peroxide as radical methyl donor.

A copper-catalyzed radical methylation/sp(3) C-H amination/oxidation reaction for the facile synthesis of quinazolinone was developed. In this cascade reaction, dicumyl peroxide acts not only as a useful oxidant but also as an efficient methyl source. Notably, a methyl radical, generated from peroxide, was confirmed by electron paramagnetic resonance for the first time.

Herein, we report a new pyridine-biquinoline-derivative fluorophore L for effectively sensing pyrophosphate (PPi) and monohydrogen sulfide (HS(-)) in aqueous buffer and in living cells. L could selectively coordinate with metal ions (M(n+)) in Groups IB and IIB to form L-M(n+) complexes with 1:1 stoichiometry, resulting in fluorescence quenching via photoinduced electron transfer (PET) mechanism. L-Zn(2+) complex was applied to competitively coordinate with PPi to form a new "ate"-type complex, turning on the fluorescence by a 21-fold-increase. The limit of detection (LOD) of this assay for PPi detection in aqueous buffer is 0.85 μM. L-Cu(2+) complex was applied for highly selective detection of HS(-) with an excellent sensitivity by 25-fold decomplexation-induced fluorescence increase. LOD of L-Cu(2+) complex for HS(-) detection in aqueous buffer is 2.24 μM. With the in vitro data obtained, we successfully applied these two complexes for sequential imaging Zn(2+) and PPi, Cu(2+) and HS(-) in living cells, respectively. Since PPi and HS(-) occur in vascular calcification in positive correlation, our multifunctional probe L might help doctors to more precisely diagnose this disease in vivo in the future. For example, we could use radioactive tracer L-(64)Cu for qualitative and quantitative positron emission tomography/computed tomography (PET/CT) imaging of HS(-) in vivo.

Tetraphenyladamantane-based microporous polyimide was synthesized. It can uptake 14.6 wt% CO2 at 273 K and 1 bar, 99.2 wt% benzene and 59.7 wt% cyclohexane at 298 K and 0.9 bar, exhibiting potential applications in gas storage and recovery of organic pollutants.

We design a novel kind of polarization beam splitter based on a gold-filled dual-core photonic crystal fiber (DC-PCF). Owing to filling with two gold wires in this DC-PCF, its coupling characteristics can be changed greatly by the second-order surface plasmon polariton (SPP) and the resonant coupling between the surface plasmon modes and the fiber-core guided modes can enhance the directional power transfer in the two fiber-cores. Numerical results by using the finite element method show the extinction ratio at the wavethlengths of 1.327 μm and 1.55 μm can reach −58 dB and −60 dB and the bandwidths as the extinction ratio better than −12 dB are about 54 nm and 47 nm, respectively. Compared with the gold-unfilled DC-PCF, a 1.746-mm-long gold-filled DC-PCF is better applied to the polarization beam splitter in the two communication bands of λ = 1.327 μm and 1.55 μm.

A novel design of Au-filled photonic crystal fiber (PCF) with square lattice has been proposed in this paper. The resonance strength of the surface plasmon mode and the impacts of structural parameters of the PCF on the polarization filter characteristics are studied through the finite element method. Numerical results show that the sizes of Au wires and the symmetry of the air holes near the fiber core have a great effect on the polarization filter characteristics. In the optimization process, it was found that the resonance strengths can reach 279.10 and 399.18 dB/cm at wavelengths of 1.02 μm and 1.55 μm, respectively, which can be applied in many polarization filter devices.

Four new polyimide networks were prepared via a one-step polycondensation method by using tris(4-aminophenyl)amine and 1,3,5-tris(4-aminophenyl)benzene to react with pyromellitic dianhydride (PMDA) and naphthalene-1,4,5,8- tetracarboxylic dianhydride (NTDA), respectively. It was found that the topological structures and pore morphologies of the networks were tightly related to the geometrical shape of the net nodes, the molecular volume of the connecting struts, the reactivity of the monomers as well as the polymerization conditions. The results revealed that the reactions between triangular triamines and rod-like PMDA tended to form six-membered ring units, which were then extended to yield polyimide crosslinking networks, reflected by the complete disappearance of the endgroup signals in the FTIR and 13C CP MAS spectra, the significantly high thermal stability, large accessible surface area (818 cm2 g−1) and narrow pore size distribution centered at 4–8 A. On the contrary, the low reactivity and steric bulky naphthyl group of the NTDA monomer led to the generation of branched or hyperbranched oligomers, which were further mutually crosslinked at a high temperature to produce the polyimde networks. In this way, a small amount of unreacted groups unavoidably remained, and the polymers exhibited a heterogeneous topological architecture, a broad pore size distribution and a low surface area of only 232 cm2 g−1. The influences of synthetic control and monomer structures on the pore morphologies and properties are investigated.